Frost Dates & Planting Calendar by ZIP Code

The date every planting guide hands you is a coin flip, and almost nobody says so out loud. Your “average last frost” is the date the last freeze has a 50% chance of still being ahead of you. Half the time it is not over. I have lost a flat of tomatoes to that sentence more than once, always in the same week of May, always because a date on a chart told me it was safe. So I went and pulled the number nobody prints next to it: the date you are actually 90 percent safe, station by station, plus the same treatment on the fall end. Every date comes from the NOAA 1991–2020 U.S. Climate Normals for the reporting station nearest you. Nothing is gated, nothing is paywalled, and the whole 365-station dataset is downloadable below.

The number nobody publishes: across the 356 stations that publish the full set of freeze probabilities, the date when you are 90% safe from a spring freeze falls a median of 16 days after the “average last frost date” everyone quotes (mean 17.6 days; range 9–39). At the other end, the date when fall frost already carries a 10% chance arrives a median of 15 days before the average first frost. Put together, planting to the averages instead of the safe window costs a median of 30 days of frost-free certainty. And the gap is widest exactly where gardeners are told they can plant early: the longer your growing season, the bigger the gap, so the mild-winter stations with long ragged freeze tails — North Bend OR and Crescent City CA both give up 66 days, Brownsville TX 62 — are the least trustworthy places to plant on an average date, while Anchorage gives up only 9 days in spring. The average last frost date is, by definition, the date your last frost has a 50% chance of still being ahead of you. It is a coin flip, and almost every planting guide treats it as a green light!

How this tool works

Now, two data layers, both public and both cited.

Layer 1 — frost dates. NOAA’s National Centers for Environmental Information publishes the 1991–2020 U.S. Climate Normals, which include freeze/frost probability dates for every qualifying station. This tool takes four published values per station, all at the 32 °F freeze mark: the last spring freeze at 50% and at 10% probability, and the first fall freeze at 50% and at 10%. Nothing is averaged or adjusted in between. Each one is a freeze-probability field published in the Climate Normals themselves, linked in the sources at the end, so you can pull the same four numbers for any station yourself. 365 stations are loaded, covering all 50 states and the District of Columbia.

Layer 2 — planting offsets. Every planting offset in the calendar is a published Cooperative Extension recommendation expressed in weeks relative to the last spring frost, and every indoor-sowing date is a published transplant age — the number of weeks extension services say a seedling needs between sowing and going in the ground — counted back from the earliest set-out date. Anchoring transplant age to set-out rather than to the frost date is what keeps cold-hardy crops coherent: cabbage goes out four to five weeks before the last frost, so a frost-anchored indoor date would have you sowing seed after the plants were already in the bed. Sources are named per crop in the reference table below. Where extension services disagree, the table reports a representative published range rather than an average, and the disagreement is disclosed.

Location matching. Your ZIP code’s first three digits are matched to a land-area-weighted centroid from the U.S. Census ZCTA gazetteer, then to the nearest of the 365 NOAA stations by great-circle distance. Across all 891 covered ZIP prefixes the median distance to the matched station is 23 miles; 90% are within 60 miles; the worst case is 135 miles. The tool always shows you which station it used and how far away it is, so you can judge the match yourself.

The formula, in plain text

No modeling, no smoothing, no adjustment factors. Every output is one of your station’s four published NOAA dates plus or minus a published extension offset.

Worked example

ZIP 64108 (Kansas City, Missouri), season year 2027, average basis. The tool matches ZIP prefix 641 to Kansas City International Airport (USW00003947), 14 miles away.

• Average last spring freeze: April 12, 2027. 90% safe date: April 27, 2027.
• Average first fall freeze: October 23, 2027. 10% risk date: October 6, 2027.
• Frost-free season: 194 days on the averages, 162 days on the conservative window — a 32-day difference.
• Tomato (6–8 week transplant age, set out 0 to +2 weeks): set out Apr 12–Apr 26, so start seed indoors Feb 15–Mar 1. Switch to the safe basis and set-out moves to Apr 27–May 11, indoor sowing to Mar 2–Mar 16.
• Broccoli (5–7 week transplant age, set out −4 to −2 weeks): set out Mar 15–Mar 29, start indoors Jan 25–Feb 8 — which is exactly what University of Maryland Extension publishes: five-to-seven-week-old transplants set out four weeks before the last frost date.
• Peas (direct-sow, −6 to −2 weeks): sow Mar 1–Mar 29.
• Garlic (4–6 weeks before first fall frost): plant cloves Sep 11–Sep 25, 2027.
• Latest possible fall sowing for a 90-day tomato: Jul 25, 2027 — the bare arithmetic, before any allowance for shortening days.

Frost dates for 33 major U.S. cities

The full 365-station dataset is downloadable at the end of this page. The 90% safe and 10% risk columns are the ones almost no other frost-date page publishes.

Published planting offsets by crop

The calendar above is built from a published transplant age and a published set-out offset for each of 31 crops, every figure attributed to the extension service that published it. That full table — transplant age, set-out window, days to maturity, minimum soil temperature and sources per crop — now lives on its own page, along with the two-step method for counting back from set-out rather than from the frost date, which is what keeps cold-hardy crops from coming out backwards.

When to start seeds indoors: transplant age by crop

Frost, freeze and hard freeze: what the words mean

This matters because the NOAA dates this tool uses are freeze dates at 32 °F, not frost dates. Visible frost can settle on your beds at an air temperature of 35 °F when the sky is clear and the air is still, because the thermometer sits about five feet above ground and your seedlings do not. If anything, the dates on this page are optimistic for a low-lying garden.

How to read your results

Your last frost date is a 50/50 bet, not a deadline

Here is the single most common mistake in home gardening: treating “last frost: April 12” as permission to plant tomatoes on April 12. By NOAA’s own definition, in half of all years the last freeze arrives after that date. If losing a flat of tomatoes matters to you, use the 90% safe date; if you are prepared to cover plants or replant, the average date buys you two to three extra weeks of season.

Airport stations run warm

Remember that nearly every station in this dataset is at an airport — that is where the long, continuous, quality-controlled records live. Airports sit on open flat ground, often near pavement, and their sensors are five feet up. A rural garden in a valley bottom a few miles away routinely frosts one to three weeks outside the airport’s window. Basically, nobody publishes a correction factor for this that survives scrutiny, so this tool does not invent one. What it does is tell you which station it used and how far away it is, so you can apply your own local knowledge.

Elevation and cold air drainage beat distance

If your matched station is 40 miles away but 2,000 feet lower, the match is worse than the mileage suggests. Cold air pools in hollows on clear calm nights; a garden at the bottom of a slope can be several degrees colder than one halfway up the same hill. Use the station distance as a starting point, then adjust from what your own thermometer and your neighbors tell you.

The fall side of the last frost date is the one people forget

Spring gets all the attention. The fall side is where a season actually gets lost, and almost nobody plans for it. The “latest fall sowing” column in your results is bare arithmetic — fall anchor minus the crop’s longest published days to maturity. It assumes summer growth rates right through October, which is not how autumn works. Extension guides consistently advise sowing earlier than that arithmetic date because day length and temperature decline. Treat it as the outer wall, not the plan.

Final Thoughts

Use the 90 percent date for anything you would be sick about losing, and the average date for anything you can replant in a fortnight. That is the whole decision, and it is worth two or three weeks of season either way. If you are working out when to sow rather than when to set out, the transplant age table is the other half of it — because the six-weeks-before-last-frost rule everyone repeats is only true for a tomato.

Related on Current Homesteading

• May planting calendar by USDA zone — what the frost dates mean for what goes in the ground
• Hardening off seedlings: the 10-day schedule
• Goat and livestock gestation calculator
• Raised bed soil calculator
• Chicken coop size calculator

Frequently asked questions