What the Early, Typical and Late Frost Dates Actually Mean

A range because the real date moves every year

NOAA's 1991-2020 record does not have one "last frost date," it has 30 of them, one per year, and they land on different days. The median date on every page here is the middle of that spread, useful as a planning anchor, misleading if treated as a guarantee.

The safe date and the risky date

Each ZIP page also gives the two ends of that spread: a date by which frost has occurred in roughly 9 years out of 10 (still risky to have planted before), and a date by which frost has occurred in roughly 1 year out of 10 (a safer cutoff to wait for). Planting tender crops right at the median date is a real coin flip against the last three decades of local weather, not a rule.

A naming trap in the source data worth knowing about

NOAA's own column names for these dates ("10% probability," "90% probability") describe a probability threshold, not which calendar date is earlier. For the spring last-frost event specifically, the "10%" figure turns out to be the later, safer date and the "90%" figure the earlier, riskier one, the reverse of what the names suggest at a glance. This site compares the actual dates rather than trusting the column names, so the "risky" and "safer" labels on every page are always chronologically correct.

What this does not cover

A frost date normal cannot see this year's actual weather, or a late-season cold snap after an early warm spell. Growers who plant right at the risky end of the range typically hedge with row covers or cold frames rather than betting on the average alone. See the microclimates guide for what else moves the real date at your own address.

Where to go next

See the range this plays out across: Stanley, ID, Farber, MO, Indian Creek, FL.

Why Your Own Yard Might Not Match Your ZIP Code's Number · How FrostDateCheck Calculates Your ZIP Code's Frost Dates