Your Last Frost Date Is a Probability
Your "last frost date" is not the day frost stops. In most planting calendars it is the median date of the last spring freeze over roughly 30 years of records, which means there is still about a 50% chance of freezing weather after it. NOAA's climate data gives dates at several probability levels, and choosing among them is a decision about how much risk you want to take with your transplants.
Where frost dates come from
Frost dates in the U.S. come from NOAA's National Centers for Environmental Information (NCEI). The current U.S. Climate Normals are 30-year summaries for 1991–2020, drawn from more than 7,300 temperature-reporting stations. They include "growing season length for various temperature thresholds" along with hundreds of other statistics.
NOAA's Climate.gov map of the average last spring freeze shows, for each station, "the average date on which the chances of freezing temperatures drop below 50 percent." The page is direct about what that means: the last freeze "varies from year to year based on the weather, which means we never know the exact date in any given year until well after it has passed."
So when a website or seed packet gives you one date for your ZIP code, it is almost always this 50% figure. Some sites don't say which probability they use at all.
Reading a freeze probability table
NC State Extension published a clear guide to interpreting NCEI freeze/frost probabilities. The tables have two dimensions:
- Temperature threshold. The tables step down in 4°F increments: 36°F (frost), 32°F (freeze), 28°F (hard freeze), then 24°F, 20°F and 16°F.
- Probability level. Nine levels, from 10% to 90%.
For spring, each cell reads as "there is an X% chance of reaching this temperature on or after this date." Here is NC State's example for a Raleigh station at the 36°F frost threshold:
| Probability | Spring frost (36°F) on or after | How often you'd be wrong | Risk profile |
|---|---|---|---|
| 90% | March 28 | Frost comes later in 9 of 10 years | Aggressive |
| 50% | April 12 | Frost comes later in 5 of 10 years | Moderate |
| 10% | April 26 | Frost comes later in 1 of 10 years | Conservative |
Four weeks separate the aggressive date and the conservative one at a single station. That spread is typical of how variable spring weather is, and it is the information a single "last frost date" throws away.
Fall works the same way in reverse. NC State's Raleigh example for a 28°F hard freeze gives November 7 at 10%, November 25 at 50% and December 14 at 90%, each read as "on or before" that date.
One practical wrinkle: NC State notes that the station tables it describes were based on 1981–2010 data, with updates to the 1991–2020 period expected. Check which period your source uses.
Choosing a threshold and a probability
You pick two things: the temperature that actually hurts your crop, and how often you're willing to lose.
Temperature. For frost-tender crops like tomatoes, peppers, squash and basil, the 32°F or even 36°F column is the relevant one. For hardy crops like peas, spinach and cabbage, a 28°F date is often more useful. The 36°F column is there for a reason. Official temperatures are measured about five feet above the ground, and the National Weather Service explains that on clear, calm nights the air near the ground can be "several degrees colder." A night that officially bottoms out at 35°F can still put frost on your seedlings.
Probability. NC State maps the levels to risk tolerance:
- 90% date: risk-seeking. Fine for cheap direct-sown crops you can resow.
- 50% date: moderate. You'll need to watch forecasts and cover plants some years.
- 10% date: risk-averse. Good for transplants you spent two months raising.
A reasonable approach for a home garden is to use the 10% date at 32°F for tender transplants, and to keep a row cover ready in case of the one-in-ten year.
Your yard is not the weather station
Station data describes the station. Your yard can run colder or warmer, and the difference is biggest on exactly the nights that matter.
Frost damage in spring usually comes from radiation frosts. Auburn University's freeze protection guide describes these as calm nights, usually 0 to 3 mph wind, with clear skies. Heat radiates from the ground into space and cold air settles at the surface. The guide notes that on those nights, temperatures at the soil surface or on low crop beds "may easily be several degrees colder" than at five feet. The NWS adds that valleys are more prone to frost because cold, dense air sinks and collects there.
Wind-driven freezes are different. Auburn describes them as arctic air masses arriving with winds usually above 5 mph, and notes that protection methods are of "limited value" against them. Row covers can help with a calm radiation frost and do much less against a windy hard freeze.
What this means in practice:
- A low spot at the bottom of a slope will frost more often than the station suggests.
- A bed against a south-facing wall may run warmer.
- Clear, still nights with forecast lows in the mid-30s are frost risks for tender plants, even if the forecast never says 32°F.
The climate is shifting under the averages
Normals are backward-looking by design. The EPA's growing season indicator finds that the frost-free season in the contiguous 48 states has grown by more than two weeks since the beginning of the 20th century. Compared with the long-term average, the last spring frost has come about 3.5 days earlier and the first fall frost about 3 days later since 1980. Western states have changed faster than eastern ones, and the EPA notes that Alabama and Georgia saw their growing seasons get shorter.
An earlier average does not mean fewer bad years. A trend of a few days is small next to the four-week spread in a single station's probability table. It does not justify planting tomatoes early.
Finding your numbers
- Open NCEI's Climate Normals tools and find the station nearest you with a similar elevation and setting.
- Look for the spring freeze probability values at 36°F, 32°F and 28°F.
- Write down the 10%, 50% and 90% dates for each threshold you care about.
- Match them to crops: tender transplants to the conservative 32°F date, hardy crops to earlier ones.
- Check your soil too. A frost-free date does not mean the soil is warm enough. See Soil Temperature for Germinating Vegetable Seeds.
If the raw data is hard to navigate, check whether your state extension service has a guide. NC State's covers 120 North Carolina stations and is a good model for reading any state's tables.
Once you have your dates, count back to indoor sowing and plan a week or two of hardening off before the transplant date. If you want to gamble on an earlier planting, row covers and cold frames buy a few degrees. For the full spring sequence, go back to Seed Starting and Season Extension.