Frost dates: what the date means, and what to do tonight

Every gardener is told to wait a couple of weeks past the frost date to be safe. Nobody tells you what the wait costs — or that on a clear night a forecast of 38 already means frost.

What does a "last frost date" actually mean?

It's a bet, quoted at even money. The date on the seed packet comes from thirty years of records at a weather station near you, and it marks the day after which half of those years saw no more frost. The other half did. It isn't a deadline the weather has agreed to, and treating it as one is the single most expensive misunderstanding in vegetable gardening.

Three dates, not one

NOAA's freeze normals give the date with a 90%, 50% and 10% chance of frost still to come. Almanacs and packets print the middle one. Plant tender crops on the 50% date every spring and — over a gardening lifetime — you lose them about 5 years in every ten. Wait for the 10% date, roughly a fortnight later, and it's 1 year in ten. Neither is "the" frost date; they're two points on the same curve, and choosing between them is a question about your appetite for replanting, not about the climate.

What waiting two more weeks actually costs

Here's the part every "just wait to be safe" article leaves out. Safety comes out of the front of the growing season, and the growing season is not the same length everywhere. The same fortnight of caution is a very different purchase in Minnesota than in Georgia:

ZoneFrost-free seasonWaiting 14 days leavesShare of the season given up
3123 days109 days11.4%
4136 days122 days10.3%
5166 days152 days8.4%
6186 days172 days7.5%
7213 days199 days6.6%
8240 days226 days5.8%
9284 days270 days4.9%
10318 days304 days4.4%

Season length between the published spring and fall dates for each zone's typical frost dates; the wait is 14 days, the usual gap between the 50% and 10% thresholds. Zone dates are approximations — a station lookup beats a zone every time.

Caution costs the most exactly where seasons are shortest

In zone 3 the season is only 123 days, so a fortnight of waiting spends 11.4% of the whole year's growing. In zone 9 — 284 days — the identical wait costs 4.9%. The gardener who can least afford to lose two weeks is the one most often told to wait, and the gardener with time to spare is the one who rushes.

Which is why the useful move isn't picking a braver date — it's making the wait cost nothing. Start the long crops indoors so they're going out as transplants (the seed-starting calculator counts the weeks back for you), and keep a row cover in the shed so a forecast doesn't force a decision. Both buy the 10% date's safety at the 50% date's speed, which no amount of staring at the calendar can.

Your zone's frost window

The published date with a fortnight either side — the range a real year lands somewhere inside. Read the late spring column and the early fall column together: that's the season you can count on nine years in ten, and it's meaningfully shorter than the one on the seed packet.

ZoneLast spring frostFirst fall frostTypical season
Early (risky)PublishedLate (safe)Early (safe)PublishedLate
3May 1May 15May 29Sep 1Sep 15Sep 29123 days
4Apr 28May 12May 26Sep 11Sep 25Oct 9136 days
5Apr 16Apr 30May 14Sep 29Oct 13Oct 27166 days
6Apr 3Apr 17May 1Oct 6Oct 20Nov 3186 days
7Mar 20Apr 3Apr 17Oct 19Nov 2Nov 16213 days
8Mar 6Mar 20Apr 3Nov 1Nov 15Nov 29240 days
9Feb 6Feb 20Mar 6Nov 17Dec 1Dec 15284 days
10Jan 17Jan 31Feb 14Dec 1Dec 15Dec 29318 days

Zone frost dates are approximations — hardiness zones are defined by winter minimum temperature, not by frost timing, so two places in zone 6 can differ by weeks. The ±14 days is the typical 50%-to-10% gap at US stations; your own station's normals are the real answer and they're free from NOAA.

The forecast says 38 tonight. Do I cover?

Probably, and this is the question the frost-date articles never get to. A forecast low is a thermometer in a shelter, five feet off the ground. On a clear, still night the ground and the leaf surfaces radiate their heat straight out to the sky and end up 2–5 °F colder than the number on the forecast. That's why gardeners get caught: the app said 38, and the basil still died at 33 °F on the leaf.

Forecast lowClear and calm — leaves reachClear and calmCloudy or breezyWhat to do on the clear night
44 °F39 °FNo frostNo frostNothing to do.
40 °F35 °FPatchy frostNo frostCover the tender things. Low spots and open ground go first.
38 °F33 °FPatchy frostNo frostCover the tender things. Low spots and open ground go first.
36 °F31 °FFreezePatchy frostCover, or pick. Uncovered tomatoes, peppers, basil and squash are done.
33 °F28 °FHard freezePatchy frostA sheet will not save tender plants. Harvest what you can; mulch the perennials.
30 °F25 °FHard freezeFreezeA sheet will not save tender plants. Harvest what you can; mulch the perennials.

Leaf temperature is the forecast low less the 5 °F worst case for a clear, calm night; cloud cover or wind mixes the air and largely removes the effect. Thresholds are the National Weather Service's: patchy frost at 36 °F and below, freeze at 32, hard freeze at 28.

The practical shorthand: clear, calm and under 40 means cover something. Wind and cloud are your friends on a cold night — they stir the warm air back down and stop the ground radiating to open sky. A still, cloudless, dead-calm night in the high thirties is exactly the night that kills tomatoes.

Frost, freeze, hard freeze: the words mean different things

TermAt or belowWhat it does
Frost advisory36 °FPatchy frost forming on surfaces. Tender growth burns; hardy plants shrug.
Freeze warning32 °FWater freezes in the leaf. Tomatoes, peppers, basil and squash are finished.
Hard freeze28 °FThe freeze reaches deeper tissue and lasts. Even half-hardy plants die back.

National Weather Service alert thresholds. Frost can form above freezing on clear nights — see the leaf temperature above — which is why a "frost advisory" starts at 36 °F rather than 32.

What survives which is mostly about the plant, and the split is worth knowing before you spend an evening carrying sheets around. Hardy — kale, spinach, garlic, most perennials, cool lawn grasses — take a freeze and often taste better for it. Half-hardy — lettuce, chard, carrots, potatoes above ground — take a light frost and die in a hard freeze. Tender — tomatoes, peppers, basil, squash, beans, cucumbers — are finished at 32, and damaged before it. Cover the third group. Ignore the first.

What actually protects a plant

A cover works by trapping the heat the ground is radiating, not by insulating the plant — which is why every part of the how matters more than the what. Put it on before dusk, while there's still ground heat to trap; a sheet thrown over frozen plants at 10 p.m. is decoration. Take it to the soil on all sides, so the warmth can't escape sideways. Keep fabric off the leaves where you can — hoops, stakes, a tomato cage — because anything touching a leaf conducts the cold straight into it, which is why plastic sheeting alone is the worst option and old bedsheets and row cover are the best. Uncover in the morning, or a sunny day cooks what the night spared.

Watering the bed the afternoon before genuinely helps: wet soil holds several times the heat of dry soil and releases it overnight. The upturned-bucket trick works for the same reason on individual plants. What doesn't work is a cover with a gap at the bottom, or hoping.

Your yard is not the weather station

Cold air behaves like water: it flows downhill and pools. A bed at the bottom of a slope, in a hollow, or behind a solid fence that dams the drainage can run several degrees colder than one twenty feet uphill — enough to shift its effective frost date by a week or two in either direction. South-facing walls, paved driveways and the shelter of a building do the opposite, radiating stored heat all night; that's why the tomato against the house survives the night the one in the open bed doesn't. A city runs warmer than the countryside it sits in, sometimes by a full zone.

None of this is in your zone's number, and all of it is knowable from two seasons of paying attention. A $15 min/max thermometer left in the bed for a spring tells you more about when to plant than any published date will.

The fall date decides whether a crop finishes

Spring frost dates get all the attention because that's when everyone's impatient, but the fall date is the one that decides whether a crop ripens at all. Days-to-maturity on a seed packet is counted in good growing weather, and the last few weeks before frost aren't that — light is short, nights are cold and everything slows. Take the early fall column above as your working date, not the published one, and give a long-season crop a couple of weeks of slack on top.

Counting backward from that date — for a fall planting, or to know the last honest day to sow — is arithmetic the seed-starting calculator already does per crop, in both directions. For lawn work, where the fall window sets overseeding and the last feed, the planner shifts the whole calendar to your zone.

So which date should you use?

Use the published date to plan — to time seed starting, order seed, and know roughly when the season opens — and use the forecast to act. The date tells you what week to be ready; the ten-day forecast tells you which night to plant, and the sky that evening tells you whether to cover. Gardeners who lose plants are almost never the ones who used the wrong date. They're the ones who used a date at all, on the night the forecast said 38 and the sky was clear.

Sources: NOAA/NCEI U.S. Climate Normals, freeze/frost occurrence probabilities (the 10% / 50% / 90% thresholds and the typical spread between them); National Weather Service frost advisory, freeze warning and hard freeze criteria; NWS and state climate office guidance on radiational cooling and why sheltered air temperature over-reads surface minimums. Zone frost dates as used across this site are approximations published by seed companies and almanacs — hardiness zones are defined by winter minimum temperature, not frost timing.