When to Start Seeds Indoors: Count Back From Your Own Last Frost

It all hangs on one date

Nearly every seed-starting instruction you'll ever read is built the same way: start 6 to 8 weeks before your last frost. Then it stops there, because the frost date is different for everybody reading and nobody can guess yours. So what you're holding is an instruction with a blank in the middle of it. The blank is the whole job.

Here's how far apart people really are. In Phoenix the typical last frost is January 5, and in Denver it's May 5, which is four months between two American gardeners reading the same line on the back of the same seed packet. A calendar that says "start tomatoes in March" is right for a thin band of the country. It's wrong for everybody else.

So we'll do this in the order that works. You get your own frost date first, because every other date on this page is counted back from it, then you do the subtraction, and then you've got a day you can write on a seed packet in pen. That's the method.

One more thing before the table. A last frost date is an average, not a promise. It's the date the freeze is finished by in about half of years, which leaves the other half of years where it turns up later than that and catches somebody's tomatoes out on the step. That's the number working correctly, which surprises people. What to do about it is at the bottom of this page.

Your typical last frost, and what it means for tomatoes

These are median last-freeze dates out of NOAA's 1991–2020 climate normals: the date a 32°F freeze is finished by in half of years, measured at each city's long-record weather station. Every line below sits on 25 to 30 years of records.

The third column does the tomato arithmetic for you, because tomatoes are what most people are counting back for. Six to eight weeks before whatever's in column two, worked out for all 28 cities. That's all it is.

How many weeks back, crop by crop

Now subtract. Every start-indoors range in the table comes from what a university extension service publishes for that crop, and the citations are down at the foot of the page. Where we don't hold a sourced date for the move-outside column, the cell says so instead of guessing. We'd rather leave a gap.

Watch the third column, because that's the one that catches people out. A few of these crops go outside before the last frost has even happened. Broccoli, cabbage and collards want the cool end of spring and will sit through a light freeze without complaining. Tomatoes won't.

Counting it out, start to finish

Three steps, one crop at a time. Do it once on paper and you won't have to think about the calendar again until the trays go outside.

1. Get your last frost date. Take it off the table above if your city's up there, or run your own address through the free frost date lookup. Then write the date somewhere you'll find it again in February, because every other date this season gets counted back from that one.
2. Subtract the weeks for that crop. Find the crop in the chart and count backwards on a calendar. Tomatoes at 6 to 8 weeks back from a May 5 last frost land you on March 10 to March 24. Any day in that range works. Nobody's holding you to March 10.
3. Write the sowing date on the packet. Write the date straight on the packet in pen, then stack the packets in date order in the drawer. Come January you won't redo any of this arithmetic, and you won't miss the onions while you're sitting there waiting on the tomatoes.

Do the slow ones first. Celery and artichokes want 8 to 12 weeks and onions want 8, so all three get sown while the tomato packets are still in the drawer, and they're the ones people forget about until it's too late to do anything with them.

Do not start these indoors at all

Some crops come out worse for the trouble. These go straight in the ground, and the tray does nothing for them:

Carrots, radishes, parsnips and beets. The part you're going to eat is the root. Knock it about while it's forming and you'll pull up something forked and twisted instead. A root gets one go at being straight.

Beans, peas and corn. They come up fast and they hate having their roots messed with. A bean sown straight in the ground will overtake a transplanted one inside two weeks.

Cucumbers and squash. You can start these indoors, and plenty of people do, mostly to buy two or three weeks back in a short season. Which is fair enough. If your season is comfortable, though, sowing them straight in the ground gets you a better plant for less work, so this one depends on where you live more than the other two do.

There's a pattern under all of that, and it takes two things at once. If the seed is big and it comes up fast, put it in the ground. The tray is for the small fussy seed that comes up slowly and needs a head start it can't get outdoors in cold soil.

Started too early? You are fine

This is the most common seed-starting mistake there is, and it's nowhere near fatal. February turns nice for a week, the sun comes through the kitchen window like it means it, and the tomatoes go in six weeks ahead of schedule.

What you get is a leggy seedling: tall, thin, pale, leaning hard at the glass. That's the plant reaching for light it isn't getting, in a pot it outgrew a while back, waiting on weather that isn't ready for it. All three at once.

The fixes, in order of how much they help. Get the light closer. A shop light hung a couple of inches above the leaves does more than the sunniest windowsill you own. Pot the plant up into something bigger so it isn't root-bound while it waits. And with tomatoes, bury the stem deep when you transplant. They root all along the buried stem, which means a leggy tomato planted deep ends up stronger than a stocky one planted shallow.

And if it's genuinely too far gone, go buy a transplant. Nobody's keeping score. A four-dollar plant off a nursery bench in May beats a tray of exhausted seedlings, and the seed you didn't use keeps until next year.

Starting late is the easier problem. A tomato sown three weeks behind schedule fruits about three weeks later, and that's the whole penalty, unless your first fall frost gets there before the fruit ripens. Check that before you sow anything late.

The part the average hides

Everything above hangs off a median, which is the date frost is finished by in half of years. The other half of years, it isn't.

That barely matters on the day you sow. It matters enormously on the day you carry the trays outside. The dates here will get your timing right, and they still can't tell you whether next Tuesday night drops to 30°F, because nothing built on thirty-year averages can do that.

So the last step of seed starting happens off the calendar. You check the forecast the week you plant out, every single time, and you stay willing to go out at ten at night and carry the trays back in. That's the step that loses people whole flats of seedlings, and no chart can do it for you.

Questions people actually ask

Sources

Every planting date, depth, and temperature on this page traces to these university extension guides. Where they differ, the ranges above hold the whole spread.

1. NOAA National Centers for Environmental Information — U.S. Climate Normals 1991–2020, freeze/frost probabilities
2. Clemson Cooperative Extension — Tomato
3. Clemson Cooperative Extension — Pepper
4. Clemson Cooperative Extension — Eggplant
5. Clemson Cooperative Extension — Basil
6. Utah State University Extension — Peppers in the Garden
7. Utah State University Extension — Broccoli in the Garden
8. Utah State University Extension — Lettuce in the Garden
9. University of Maryland Extension — Growing onions in a home garden
10. University of Maryland Extension — Growing peppers in a home garden
11. University of Minnesota Extension — Growing basil