GardenTrack
Join the beta

Your frost dates, and every date built on them

Live · This works today.

Almost every date you see is calculated backwards or forwards from two numbers: your last spring frost and your first fall frost. We work those out for your own roughly 800-foot square of ground from 26 years of daily temperature records, not from the nearest airport. By default both are the average — the 50% line — and setup lets you shift them to the cautious or the early end of the record instead, which moves every date that hangs off them.

What gets calculated from them

More than most people expect:

  • When to start seeds indoors — counted back in weeks from your last frost.
  • When to plant out — counted forward in days from it.
  • When to direct sow — before or after, depending on the crop.
  • Whether a crop will finish at all — the app checks whether the harvest window opens before the frost that actually ends that crop, and tells you plainly if it does not. The same sum, run forwards, is the "what can I plant right now?" list on the add-a-plant screen. It took three goes to stop being confidently wrong, a fourth to stop being too generous about August, and a fifth to stop using one frost date for every crop — see below.
  • Every stage in between — sprouting, leafing out, flowering, fruiting. The whole timeline hangs off these two dates and the day you actually planted.
  • When to expect your seeds up — a plant you sowed carries a line saying roughly how many more days until the first sprouts, and it says plainly when the window has passed so you can decide about re-sowing. Log that it germinated and the line goes away — it briefly did not, and kept counting down at people who had just told it they had seedlings. Not every crop gets the line. It needs a first day and a last day, and for about a quarter of our crops we only have one of the two. Wisconsin says a cucamelon comes up "within 10 days", which is a last day with no first day. We would rather show nothing than invent the missing half, so those plants simply do not carry the countdown.

An August day is not worth a June day

A seed packet says "60 days" and means 60 spring days — lengthening light, warming soil, a season working with the plant. Count those same 60 days forward from August and the sum is clean but the plant does not cooperate: it is growing into shortening days and cooling nights, and it takes longer.

So for anything planted after midsummer we add two weeks to every crop, and another two weeks on top for anything a frost kills outright. Both numbers come from Nebraska Extension's fall vegetable guide, which calls them the Fall Factor and the Frost Tender Factor. Switching them on at midsummer is our reading of their rule, not a date they publish.

It changes real answers. On 3 August, with a first frost around 13 October, a tomato start went from "16 days to spare" to "not this year" — which is what any gardener would have told you, and what the app should have said.

A frost does not end everything

A tomato dies at the first light frost. Spinach does not — it barely notices, and a light frost actually makes kale and a few others taste better. Until 7 August 2026 the app judged every crop against the same date, the first light frost, and that told people it was too late to sow spinach, lettuce, radish, kale, turnips and other hardy crops weeks before it actually was. That was wrong, and it cost people plantings they had time to make.

Now a crop that a light frost does not kill is checked against your first hard freeze instead — a colder, later date we already track for every address, the same careful way as your other frost dates. A crop that really does die at any frost, like a tomato or a pepper, still uses the light-frost date; that part was never wrong.

The difference is real. In Fargo, ND — about as short a growing season as the Lower 48 has — turnips sown on 7 August went from "not this year, three days short" to "yes, with a week to spare," because the app started checking against 12 October instead of 2 October. Most of the country has it easier than Fargo does.

Your variety, not just your crop

Our tomato record says ripening runs 52 to 90 days from transplanting, and that range is honest for tomatoes in general and wrong for every particular one. Sungold is 57 days. Ace 55 is 85. Broccoli is worse: Melody is a fast summer variety and Purple Sprouting is an overwintering one that takes the better part of a year, and they share a crop record.

So when you tell the app which variety you planted, it dates that variety. 2,224 of our varieties carry their own sourced days-to-maturity, and 1,615 of them line up with their crop's calendar closely enough to use safely. Across the catalog the spread inside a single crop runs to a month, which is more than most of the corrections we make.

Where it declines, and why. A variety's figure is counted either from sowing or from transplanting, and those differ by the whole time a seedling spends indoors — six to eight weeks on a tomato. Where the variety counts from one and our crop calendar counts from the other, we leave the crop's range alone rather than convert. A converted guess would look more precise and be less true.

If you do not know the name, scroll it. Tap the variety box on a plant and the whole list is there before you type anything. On the crops where we hold it, the list is sorted by the shape the plant grows into — vining ones that keep going until frost, bush ones that set their crop over a few weeks and stop — because that is a question you can answer standing at the plant, and "70 days" is not. Where we do not hold it, the list is simply A to Z. Rows carry what each variety is known for, in plain words: takes the heat, keeps a long time after picking, no thorns.

Your real dates beat our estimate

When you tell the app the day you actually sowed or planted something, that date wins. The frost calendar is only used to fill in what you have not told us.

This matters more than it sounds. Somebody who sowed three weeks late does not want a schedule built on when they could have sown — they want one built on what they did.

"A couple of months ago" can mean last year, and now it does. The add screen asks roughly how long ago something went in rather than making you find a date on a calendar. Answer it in January and the real date is usually in the year before — and until August 2026 the app filed it under the new year instead. A plant put in the ground in November, added on 10 January, showed up as planted on 1 January and counted as a plant from this season rather than one already into its second. For a tree or a bush that is the difference between a first-year job list and a second-year one. It was wrong for anything added between 1 January and 1 March, which is exactly seed-starting and bare-root season, and it is fixed.

An average is not a promise, and you can choose which end to trust

Unless you say otherwise, the date the app uses is the 50% line: in half of all years, frost comes later than that. The free frost lookup shows the 10% and 90% dates too, which is the honest spread — and one spot can run more than a month between those two ends.

You do not have to do that arithmetic yourself. Setup asks how much do you want to gamble, and the answer moves every date in the app: Play it safe counts against the date frost is finished by nine years out of ten, Middle of the road is the average described above, and Push my luck uses the early end for gardeners who have row cover and do not mind using it. You can change it whenever — Profile, then Walk me through setup again. In places mild enough that some winters never freeze, there is no early date to offer and the app says so rather than showing a button that does nothing.

Common questions

The dates for something I started indoors just moved. Why?
**We were working out the indoor stretch wrongly, and it was always too short — fixed 25 August 2026.** For a plant you start inside, we need to know roughly how long it spends indoors before it goes out. We hold two figures for that: how many weeks before the last frost seed is usually started, and how many days after the last frost it usually goes out. Both are ranges — a tomato is *6 to 8 weeks* in, then *7 to 14 days* after. **We were taking the shortest of each: 6 weeks and 7 days.** That pairs the two ends least likely to happen together — somebody who starts as late as they might does not also plant out as early as they might. The honest answer is the middle of each, and for a tomato that is 60 days indoors rather than 49. **Two plants were coming out as zero days indoors** — sweet pea and bok choy — which put the day they were sown and the day they went out on the same date. Of the 79 plants that use this figure, **52 move by two weeks or more.** **What you will notice:** stage dates for anything started indoors sit a little later than they did, and they are closer to right. Nothing about a plant you sowed straight into the ground has changed.
I bought my onions as plants. Why did their dates suddenly get better?
**Because from 24 August 2026 we work backwards from the day you set them out.** Most dates for a plant are counted from the day its seed was sown. But if you bought a start, or someone gave you one, you never sowed it — so the one date you can tell us is the day it went in the ground. For a long time that left us with nothing to count from, and the app fell back to a national calendar: a date about a typical onion somewhere in America, shown to you as a fact about yours. **It was not a small gap.** An onion set out ten days ago was being reported as bulbing, ready to harvest, and ready to cure — and the daily list told you to go and harvest it. That is not a rounding error. It is a completely different plant's summer. **The fix is arithmetic, not a guess.** We know how long an onion spends indoors between sowing and planting out, because it is a sourced number in our plant records — eight weeks. Count eight weeks back from the day you planted and you have the day it was sown. Every date that hangs off sowing then belongs to your plant instead of an average one. **Five crops gained this in one go: onions, cucumbers, okra, pumpkins and summer squash.** Three still cannot do it — fennel, kale and Brussels sprouts — because we do not yet hold the number for them. **Those will keep showing the typical calendar and saying so**, because putting a made-up number on your own plant is the thing this whole change exists to stop.
The harvest date on my plant moved later. Did something change?
**Yes, on 18 August 2026, and the new date is the more honest one.** A plant growing into autumn does not grow at the speed it did in June. Days shorten, soil cools, and the last stretch to a ripe fruit takes longer than the seed packet's day count suggests. The app already knew that and already allowed for it — **in one place and not the other.** The frost check used the corrected date; the harvest window on your plant page used the raw one. **On a late crop that a frost ends, the two could sit up to four weeks apart.** Same app, same plant, two answers, and the earlier one is the one that quietly tells you there is time when there might not be. **They now both use the corrected date**, so your harvest window may have shifted later by up to about four weeks on a late-season crop. Nothing about your plant changed. What changed is that we stopped showing you the optimistic number in one corner of the app and the realistic one in another. **We picked the later date deliberately.** Being told you will get fruit and then losing the crop to frost is a worse afternoon than being told you will not and getting some anyway.
My tomato's timeline says it is harvested in November, but frost comes in October. Which is right?
**Both dates were right and nothing had ever put them side by side. From 13 August 2026 it does.** Where a plant's timeline reaches a date after your first frost, and that plant is one a frost genuinely ends — a tomato, a pepper, basil — the list now says so on the first date it applies to: *Your first frost is usually around Oct 21 — before this.* **The dates stay exactly as they were.** Nothing is hidden and nothing is re-planned. **It is a comparison, not a forecast.** We are putting two numbers we already show you next to each other. We are not telling you the plant will die: a tomato under fleece in a warm autumn does perfectly well, an unusually late frost happens most years somewhere, and you know things about your own ground that we do not. What we will not do any more is hand you a harvest date in November without mentioning October. **If you have not pinned your garden, the sentence says so** — it will name a typical zone 6b frost rather than yours, because that is what it is comparing against, and it asks you to pin. That is the whole reason the wording differs between the two: **a warning built on a guess must not read like a warning built on your garden.** **Hardy plants are left alone on purpose.** A kale harvest after the first frost is completely ordinary and often sweeter for it. Those crops are measured against your first hard freeze instead, which is a colder and later date — so marking them against the light frost would be comparing them to the wrong number, not just fussing.
I hid the setup list, but "Pin your garden" is still there. Is that a bug?
**No — it is the one item we deliberately do not let you hide, and it is worth knowing why.** Everything else on that list costs you something you can see if you skip it. This one costs you something invisible. **Every journal entry records roughly where the plant should have been on that date, and that is worked out from your frost dates** — so while the app has no location for you, every entry you write is saved with that part blank. **Permanently blank**, because we never recalculate a stage once it is written down. So each day it stays unpinned quietly costs you entries that cannot be filled in later by hand. **It is one line, not a wall.** Nothing is blocked, there is no pop-up, and the rest of the setup list stays hidden exactly as you left it. Pin your garden and the line goes for good — **and the blank entries you already have get filled in at the same moment**, each one marked as worked out later so you can tell it from a stage recorded on the day.
Is it too late to plant fall crops?
For most of the country, no — and until 7 August 2026 the app sometimes said otherwise for crops like spinach, kale, lettuce and radish, which was wrong. A light frost does not end those crops, so we now check them against your first HARD freeze instead of your first light frost — a colder, later date, worked out the same precise way as everything else here. See 'A frost does not end everything' below for the real numbers.
Why does the app keep asking for my location?
Because without it every date is a national average, and a national average is wrong for almost everybody. Your last frost decides when to start seeds, when to plant out, and whether a crop has time to finish before fall. Get that one number right and everything downstream improves.
What if I skip it?
The app falls back to typical zone 6b dates and says so on screen, every time. Nothing is hidden. But those dates will be wrong for you by anything from a few days to several weeks.
I garden in Alaska, Hawaii or Puerto Rico. Do I get frost dates?
**Not yours, and the app says so rather than guessing quietly at you.** The map behind these dates is built from 26 years of daily temperature records covering the lower 48. Outside that we hold nothing for your square, so the lookup will not save and the garden stays unpinned. **Type your own instead — added 19 August 2026.** *Pin my garden*, then **I have my own records**. That is the whole fix: your dates are then real dates rather than a regional average, the setup nag stops, and — the part worth knowing — **the blank journal stamps fill themselves in.** Every entry you have already written gets re-stamped against your dates the moment you save them. Nothing you wrote while unpinned is lost. **Until you do, you get one fixed calendar — a typical zone 6b — the same one every unpinned garden uses**, and the screens built on it say it is a guess. Be blunt with yourself about the size of that gap: zone 6b puts the last frost around 17 April and the first around 21 October, which is neither Fairbanks nor San Juan. **Everything that does not hang off frost works exactly as it does for anybody else** — your plants, your zones, your photos, your picks, the whole journal.
I garden where it never freezes — Miami, the Keys, coastal southern California. Can I even set up?
**Yes, from 25 August 2026. Before that date you could not, and that was our fault.** Pin your garden in one of those places and the app used to say *"No frost record for that exact spot — it may be water. Try a point on the land nearby."* You are not on water. There was no way past it: nothing was saved, so the setup item could never be ticked and every date in the app stayed a zone 6b guess. **Why it happened.** The map we look your dates up in stores a blank for a square of ocean and the same blank for a garden that never freezes. The app read every blank as ocean. Our own records knew better the whole time — they have Miami and Key West down as never hitting freezing. **What it does now.** It tells the two apart and says which. If nothing near you has ever recorded a freeze, it says so. If frost turns up some years but not most — much of coastal southern California and south Florida — it says that instead, because those are genuinely different and the second one is the one that catches people out. Then press **Use this location anyway** and setup carries on. **You get no frost dates, because there are none to give you.** Nothing will pretend otherwise: no last-frost countdown, no "safe to plant out after" line built on a date we invented. Everything that does not hang off frost — your plants, your photos, watering, the journal, the whole rest of it — works exactly as it does for anybody else.
The app says frost is rare where I am. What am I supposed to do with that?
Treat it as a real risk you cannot put a date on. A place that freezes in, say, one year in five has no useful average last-frost date — the average of *mostly nothing* is not a number worth planting by. So we do not give you one, rather than giving you a confident date that is wrong four years out of five. **What it means in practice:** your tender plants are fine almost every year, and the year it does freeze it will not be on a schedule. If you have kept your own records, they beat anything we can model — *Pin my garden*, then **I have my own records**, and every date in the app moves onto yours.
Why an address rather than a ZIP code?
A ZIP centroid can sit four miles from your garden in a normal ZIP and eight in a rural one. In hilly ground that is weeks of difference in the last frost date, not days. The address gets us a latitude and longitude, we look up your square, and then we throw the address away.
How precise is it really?
The underlying map is built at about 250 meters — roughly an 820 by 786 foot cell — and corrected for the elevation of that cell. It is a 26-year average, so it is a climate normal, not a prediction. Unless you have told the app to play it safe or push your luck, half of all years your last frost will fall later than the date shown.
My garden is in a dip and always frosts early. Does the app know?
Not yet, and this is the honest limit. Cold air flows downhill and pools in low ground, so a garden at the bottom of a slope can frost weeks before the one forty feet up. We are working on modeling that and publishing the results as we go, including the attempts that fail.
Can I just enter my own dates?
**Yes — added 19 August 2026.** Go to *Pin my garden*, choose **I have my own records**, and type your last spring frost and your first fall frost. Kept records beat anything we can model for your ground. **Only the month and day are used**, so pick any year on the date wheel. Enter the dates you would plant against in an ordinary year, not the earliest or latest you have ever seen. **Your dates are used exactly as you typed them.** The *how much do you want to gamble* lever does not apply on top: that shifts our dates to the cautious or the early end of a 26-year spread, and there is no spread around a date you measured. Profile says they are yours rather than pretending a choice moved them. **This page said yes for weeks before it was true**, which was our mistake and not a small one. Now it is.
I added a plant in January and said it went in a couple of months ago. Which year did that record?
The right one — the previous year. This was wrong until August 2026: the app filed a November planting under 1 January of the new year, so the plant page showed the wrong date and the app treated it as planted this season rather than one already in its second year. That distinction drives the job list for trees and bushes. Anything added between 1 January and 1 March was affected. If you added established plants during that window, it is worth opening them and checking the planted date.
Two lines on the same screen wrote the date two different ways. Which one is right?
**Both were, and from 23 August 2026 there is only one of them.** On a crop page you had run out of time to sow, the app said *the spring window closed around 4 April* — and on the line directly underneath, *based on your last frost around May 4*. Same screen, same kind of thing, written two ways. **There were three ways of writing a date in the app.** Two were spelled out by hand and had quietly disagreed since the day they were written. The third — and by far the most common, in seventeen places — asked your browser which order it preferred. **That last one is why it lasted.** On a phone set to British or European settings, all three agreed and the page looked perfectly consistent. Whether the problem existed at all depended on whose phone you were holding. **There is now one way: month first — April 4.** If your phone is set outside the US you will see more dates change order than most people do, because those were the ones following your browser before. **No date moved.** Only how it is written. **One real fix came with it.** A date counted backwards past the first of January — the shape you get when you add an established plant and say it went in a couple of months ago — printed as *Jan 1* in a couple of places instead of the November day it actually was.
Did the calendar ever show me the wrong day?
For a narrow window, yes — fixed 7 August 2026. The little 'today' mark on a plant's calendar, and the 'what can I plant right now' list, both work out today's exact date with quick math. That math had a bug: for about an hour after midnight, on any day the clocks were set to daylight saving time — roughly March through November — it could land on yesterday instead of today. It never threw off a whole week or month, just that one day, and only in that hour. Today's date is now worked out the same solid way everywhere we show it, clock changes included. **A second one, in a different place, fixed 19 August 2026.** When you asked the chat about a plant, it was handed that plant's journal to read — and the dates on those entries were being worked out in London time rather than yours. If you garden west of London, anything you logged after about six in the evening reached the chat stamped with tomorrow. Nothing in your journal was wrong and nothing was saved wrong; it was only what the chat was shown. But it meant a question like *what did I water yesterday* could be answered off a set of dates you had never seen. The dates the chat reads are now your dates. **The two we had written down and not yet fixed are both done — 21 August 2026.** The planting date on a plant's own page read one day early for anyone gardening east of London, all day, every day, not just in the evening. That one mattered more than it sounds: the date box on that page starts out already filled in with it, so editing anything else on the plant and saving could write the early date back over the right one. The other was the date box on *when did it go in*, which would let you pick tomorrow if you garden west of London late in the evening — after six o'clock in Colorado, after eight on the east coast. Both now read the calendar on your own phone. **A third one, on this website rather than in the app, went with them.** Add a plant to your garden from a plant page in the evening and the *My garden* list said *added Aug 22* on the 21st. Only that one line was wrong. The plant itself was saved correctly, and the date the app records when you bring your picks across has always been worked out fresh on your own phone. **A date already saved a day early stays a day early until you correct it**, so if you garden outside North America it is worth opening a plant or two you added before today and checking the planted date.
How close are my frost dates to my actual garden?
**It depends which way you pinned it, and from 19 August 2026 the app tells you which rather than saying the same thing to everybody.** Give us a **street address** or let your phone share its **location** and your dates come from an 800-foot square around that point — close enough that your dates are yours. Give us a **ZIP** and we use the middle of that postal area, which can be a few miles from where you actually garden. In flat country that difference is small. In the mountains a few miles is a different growing season, and it is worth swapping to an address. Profile now says which of the two you have, under **Your garden**. If it says the ZIP one, the *Change where I garden* button on that card is where you sharpen it. **If you pinned your garden before this date we do not know which way you did it**, so it says the general thing rather than guessing — re-pin and it becomes specific.
I added beans for next Saturday. What happens if Saturday comes and I forget to tick it?
**From 19 August 2026 the app asks. Before that it quietly assumed you had planted them.** Adding a plant with a date ahead of today is a plan, and the app stays out of your way until the day arrives — then the job to plant it appears at the top of your list. If you tick it, that is the day the plant''s whole calendar is counted from. If the day comes and goes with nothing said, the next morning it asks plainly: *did the beans go in?* **Why it asks rather than assuming either way.** Every date we give you for that plant — when to expect it up, when it is ready, whether it beats your frost — is counted from the day it actually went in. Assume wrong and all of them are wrong, quietly, with no sign anything is off. Saying *not yet* is a complete answer and nothing is deleted; the plan waits. **It only asks about plants you dated ahead.** A plant you added with today''s date, or any day in the past, is a record of something that happened and is left alone. **And nothing else about that plant starts early.** A bed dated for next Tuesday is a plan, not a plant — no watering checks, no growth clock, nothing that assumes there is anything in the ground. Until 19 August 2026 it was asked to *feel the soil where you sowed*, on seed still in the packet. The one thing that does start ahead of the day is hardening off, which is meant to: seedlings go out for two weeks BEFORE they are planted.
My fruit tree was asked whether it had sprouted yet. It is five years old.
**Fixed 19 August 2026.** That question is for seed. A tree you bought and set out was already alive when it arrived, so there is no sprouting to watch for and never was. The app had the right rule already — the line on the plant page that says *"first sprouts in 5–10 days"* has always refused a plant you set out rather than sowed. The card on your daily list asking the same question simply never checked. Same question, two screens, one of them wired. **Why the other guards did not catch it.** The card waits until the earliest day a crop could possibly show, taken from that crop''s own sprouting window — and **every perennial we hold has no sprouting window at all**: orange, apple, lime, asparagus, rhubarb, strawberry, garlic. We deliberately do not silence the question for a crop with no window, because a missing figure is not evidence that a plant comes up instantly. That is the right call for the early end and it left no limit at all on exactly the plants that never sprout. **Seeds are unaffected**, including the awkward ones. Garlic goes in around October and may genuinely not be up by January; it is still asked.
Does the app stop me entering a date that cannot be right?
Yes — and until 18 August 2026 that check quietly stopped working every evening. If you record a sprout on a date before you sowed the seed, the app should stop you and ask which of the two dates is wrong, because a seed cannot come up before it was planted. It did that all morning and let it through after dark. The reason: it worked out the sprouting date using the clock at Greenwich, then compared it against a sowing date recorded in your own time. Those match for most of the day, but after about six in the evening in Colorado it is already tomorrow in London, so the sprouting date came out one day too high and stepped over the line it was meant to stop at. Nothing showed an error — the check simply went quiet. It now uses the same date math as the rest of the app. If you logged a sprout in the evening and the date looks wrong on the plant's page, it is worth opening it and correcting it; nothing was changed for you.

More on this

Shipped 2026-07-21. Last checked 2026-08-25. Something here wrong or out of date? Tell us — support docs going stale is a bug.