Last and First Frost Dates by City: 219 Cities (1981-2010 Data)

Last updated: 21 October 2026. Data period: 1981-2010.

This page lists the last spring frost date, the first autumn frost date and the length of the frost-free season for 219 cities in 21 countries. Every date comes from the frost grid and the cell lookup that Leaftide’s Frost Date Finder and app both use. The source is the Copernicus Climate Change Service (C3S) frost-day indicator for 1981-2010, on a grid of 0.5° cells. A frost here means a daily minimum air temperature below 0°C (32°F), measured at the standard height of 2 metres (about 6.5 feet).

For what frost dates mean and how to plan sowing around them, see the last frost date guide.

How to read the frost date tables

Each city has two last frost dates and two first frost dates.

• Last frost (typical) is the median last spring frost for 1981-2010. In half of those years, the last frost came before this date.
• Last frost (cautious) is the 90th-percentile last spring frost plus 7 days. In about 9 years in 10, the last frost came before the 90th-percentile date. Use this date before planting out frost-tender crops such as tomatoes, peppers, squash and basil.
• First frost (typical) is the median first autumn frost for 1981-2010.
• First frost (cautious) is the 10th-percentile first autumn frost minus 7 days. In about 9 years in 10, the first frost came after the 10th-percentile date.
• Frost-free days count the days from the last frost date to the first frost date, using either the typical pair or the cautious pair.
• † marks a city where the grid is a weaker guide, with a note under its table. Either the city uses a milder neighbouring grid cell, as Leaftide’s app does for a city in a valley next to high ground, or, for US and Canadian cities, a typical grid date is more than 14 days from the city’s 1981-2010 station normal, shown in brackets. A 0.5° grid cell averages all the ground in it, including hills, valleys, coast and built-up areas, while a station measures one spot. Near the station, its dates are the better guide.

Dates are resolved to 10-day periods, so a difference of a few days between nearby cities is not meaningful. Each city uses the grid cell whose centre is nearest to its coordinates, with one exception. Like Leaftide’s app, a city in a valley next to high ground uses the milder neighbouring cell, unless the city sits on the high ground itself (see the methodology below).

Last and first frost dates for US cities

Typical and cautious frost dates for 95 US cities, covering every state and the District of Columbia. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

† Phoenix: grid cell centred at 33.25°N, 112.25°W. NOAA 1981-2010 normal at Phoenix City, 0.8 km from the city centre: last frost Jan 6, first frost Jan 3, 365 frost-free days.

† Tucson: grid cell centred at 32.25°N, 110.75°W. NOAA 1981-2010 normal at Tucson International Airport, 10.4 km from the city centre: last frost Feb 12, first frost Dec 4, 293 frost-free days.

† Los Angeles: grid cell centred at 34.25°N, 118.25°W. NOAA 1981-2010 normal at Los Angeles Downtown USC, 5.4 km from the city centre: frost too rare for a median last or first frost date.

† Sacramento: grid cell centred at 38.75°N, 121.25°W. NOAA 1981-2010 normal at Sacramento 5 ESE, 7.2 km from the city centre: last frost Jan 19, first frost Dec 18, 339 frost-free days.

† Fresno: grid cell centred at 36.75°N, 119.75°W. NOAA 1981-2010 normal at Fresno Yosemite International, 5.9 km from the city centre: last frost Feb 2, first frost Dec 7, 309 frost-free days.

† Orlando: grid cell centred at 28.75°N, 81.25°W. NOAA 1981-2010 normal at Orlando Executive Airport, 4.6 km from the city centre: last frost Jan 17, first frost Jan 9, 364 frost-free days.

† Tallahassee: grid cell centred at 30.25°N, 84.25°W. NOAA 1981-2010 normal at Tallahassee Regional Airport, 8.6 km from the city centre: last frost Mar 19, first frost Nov 19, 243 frost-free days.

† Chicago: grid cell centred at 41.75°N, 87.75°W. NOAA 1981-2010 normal at Chicago University, 8.5 km from the city centre: last frost Apr 12, first frost Nov 3, 205 frost-free days.

† Baltimore: grid cell centred at 39.25°N, 76.75°W. NOAA 1981-2010 normal at Baltimore Customs House, 0.9 km from the city centre: last frost Mar 20, first frost Nov 24, 248 frost-free days.

† Las Vegas: grid cell centred at 36.25°N, 115.25°W. NOAA 1981-2010 normal at Las Vegas Air Terminal, 6.7 km from the city centre: last frost Jan 31, first frost Dec 11, 316 frost-free days.

† Reno: grid cell centred at 39.75°N, 119.75°W. NOAA 1981-2010 normal at Reno Tahoe International Airport, 6.3 km from the city centre: last frost Apr 26, first frost Oct 19, 177 frost-free days.

† Newark: grid cell centred at 40.75°N, 74.25°W. NOAA 1981-2010 normal at Newark Liberty International Airport, 5.9 km from the city centre: last frost Apr 3, first frost Nov 10, 219 frost-free days.

† Albuquerque: grid cell centred at 35.25°N, 106.75°W. NOAA 1981-2010 normal at Albuquerque International Airport, 5.7 km from the city centre: last frost Apr 7, first frost Nov 4, 210 frost-free days.

† New York City: grid cell centred at 40.75°N, 74.25°W. NOAA 1981-2010 normal at NY City Central Park, 8.9 km from the city centre: last frost Apr 1, first frost Nov 18, 230 frost-free days.

† Cleveland: grid cell centred at 41.25°N, 81.75°W. NOAA 1981-2010 normal at Cleveland Burke Lakefront Airport, 2.2 km from the city centre: last frost Apr 8, first frost Nov 11, 215 frost-free days.

† Eugene: grid cell centred at 44.25°N, 123.25°W. NOAA 1981-2010 normal at Eugene Mahlon Sweet Airport, 13.6 km from the city centre: last frost Apr 18, first frost Oct 22, 188 frost-free days.

† Philadelphia: grid cell centred at 39.75°N, 75.25°W. NOAA 1981-2010 normal at Philadelphia, 1.2 km from the city centre: last frost Mar 29, first frost Nov 22, 238 frost-free days.

† Pittsburgh: grid cell centred at 40.25°N, 79.75°W. NOAA 1981-2010 normal at Pittsburgh Allegheny Co Airport, 11.4 km from the city centre: last frost Apr 17, first frost Oct 26, 190 frost-free days.

† Charleston: grid cell centred at 32.75°N, 79.75°W. NOAA 1981-2010 normal at Downtown Charleston, 0.8 km from the city centre: last frost Feb 6, first frost Dec 24, 318 frost-free days.

† Columbia: grid cell centred at 34.25°N, 81.25°W. NOAA 1981-2010 normal at Columbia Owens Downtown Airport, 4.9 km from the city centre: last frost Mar 15, first frost Nov 21, 251 frost-free days.

† San Antonio: grid cell centred at 29.25°N, 98.25°W. NOAA 1981-2010 normal at San Antonio Stinson Municipal Airport, 9.7 km from the city centre: last frost Feb 24, first frost Nov 30, 280 frost-free days.

† Salt Lake City: the dates above are for the neighbouring grid cell centred at 40.75°N, 112.25°W, the cell Leaftide’s app uses here. The nearest cell (40.75°N, 111.75°W) has a cautious frost-free season under 120 days, and this cell’s is at least 30 days longer. The nearest cell has a typical last frost of Jun 5, a typical first frost of Sep 25 and 112 typical frost-free days. The NOAA normal at Salt Lake City International Airport (last frost Apr 21, first frost Oct 24) is closer to the neighbouring cell.

In this city, in at least half of the years 1981-2010 there was no frost after midwinter, and in at least half there was none before it, so the record gives no typical last or first frost date. Midwinter here is the 10-day period half a year from the least frosty point of the year (see the methodology below). It is not always the frostiest 10 days. The cautious dates still apply: they mark the 1-in-10-year late and early frosts, with a 1-week buffer.

Frost is rare in these cities. The record shows frost in fewer than 1 winter in 10, so there is no frost date to give.

Last and first frost dates for Canadian cities

Typical and cautious frost dates for 13 Canadian cities. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

† Vancouver: grid cell centred at 49.25°N, 123.25°W. ECCC 1981-2010 normal at Vancouver Int’l A, 7.6 km from the city centre: last frost Mar 18, first frost Nov 10, 237 frost-free days.

† Halifax: grid cell centred at 44.75°N, 63.75°W. ECCC 1981-2010 normal at Halifax Citadel, 1 km from the city centre: last frost May 1, first frost Oct 31, 182 frost-free days.

Last and first frost dates for German cities

Typical and cautious frost dates for 15 German cities. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

Last and first frost dates for Italian cities

Typical and cautious frost dates for 15 Italian cities. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

† Bolzano: the dates above are for the neighbouring grid cell centred at 45.75°N, 11.75°E, the cell Leaftide’s app uses here. The nearest cell (46.25°N, 11.25°E) has a cautious frost-free season under 120 days, and this cell’s is at least 30 days longer. The nearest cell has a typical last frost of Apr 26, a typical first frost of Oct 15 and 172 typical frost-free days. Which cell fits a garden better depends on the garden’s elevation.

In this city, in at least half of the years 1981-2010 there was no frost after midwinter, and in at least half there was none before it, so the record gives no typical last or first frost date. Midwinter here is the 10-day period half a year from the least frosty point of the year (see the methodology below). It is not always the frostiest 10 days. The cautious dates still apply: they mark the 1-in-10-year late and early frosts, with a 1-week buffer.

Frost is rare in these cities. The record shows frost in fewer than 1 winter in 10, so there is no frost date to give.

Last and first frost dates for UK and Irish cities

Typical and cautious frost dates for 23 cities in the United Kingdom and 3 in Ireland. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

† Inverness: the dates above are for the neighbouring grid cell centred at 57.75°N, 3.75°W, the cell Leaftide’s app uses here. The nearest cell (57.25°N, 4.25°W) has a cautious frost-free season under 120 days, and this cell’s is at least 30 days longer. The nearest cell has a typical last frost of May 15, a typical first frost of Oct 15 and 153 typical frost-free days. Which cell fits a garden better depends on the garden’s elevation.

Last and first frost dates for French cities

Typical and cautious frost dates for 11 French cities. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

Last and first frost dates for Spanish cities

Frost dates for 8 Spanish cities. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

In these cities, in at least half of the years 1981-2010 there was no frost after midwinter, and in at least half there was none before it, so the record gives no typical last or first frost date. Midwinter here is the 10-day period half a year from the least frosty point of the year (see the methodology below). It is not always the frostiest 10 days. The cautious dates still apply: they mark the 1-in-10-year late and early frosts, with a 1-week buffer.

Last and first frost dates for Dutch cities

Typical and cautious frost dates for 6 cities in the Netherlands. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

Frost dates for other European cities

Frost dates for 10 capitals and large cities elsewhere in Europe. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

In this city, in at least half of the years 1981-2010 there was no frost after midwinter, and in at least half there was none before it, so the record gives no typical last or first frost date. Midwinter here is the 10-day period half a year from the least frosty point of the year (see the methodology below). It is not always the frostiest 10 days. The cautious dates still apply: they mark the 1-in-10-year late and early frosts, with a 1-week buffer.

Frost is rare in this city. The record shows frost in fewer than 1 winter in 10, so there is no frost date to give.

Last and first frost dates for Australian cities

Australia is in the Southern Hemisphere, where spring runs from September to November. Most last frost dates there fall in the second half of the calendar year and most first frost dates in the first half, so the frost-free season runs across the new year. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

In these cities, in at least half of the years 1981-2010 there was no frost after midwinter, and in at least half there was none before it, so the record gives no typical last or first frost date. Midwinter here is the 10-day period half a year from the least frosty point of the year (see the methodology below). It is not always the frostiest 10 days. The cautious dates still apply: they mark the 1-in-10-year late and early frosts, with a 1-week buffer.

Frost is rare in these cities. The record shows frost in fewer than 1 winter in 10, so there is no frost date to give.

Last and first frost dates for New Zealand cities

New Zealand is in the Southern Hemisphere, where spring runs from September to November. Most last frost dates there fall in the second half of the calendar year and most first frost dates in the first half, so the frost-free season runs across the new year. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

† Queenstown: the dates above are for the neighbouring grid cell centred at 45.75°S, 168.75°E, the cell Leaftide’s app uses here. The nearest cell (45.25°S, 168.75°E) has a cautious frost-free season under 120 days, and this cell’s is at least 30 days longer. The nearest cell has a typical last frost of Dec 26, a typical first frost of Mar 6 and 70 typical frost-free days. Which cell fits a garden better depends on the garden’s elevation.

Frost is rare in these cities. The record shows frost in fewer than 1 winter in 10, so there is no frost date to give.

Frost dates for South African cities

South Africa is in the Southern Hemisphere, where spring runs from September to November. Johannesburg’s last frost date falls in late winter, and its first frost date falls in early winter. Source: Copernicus (C3S) frost-day record, 1981-2010, 0.5° grid.

In this city, in at least half of the years 1981-2010 there was no frost after midwinter, and in at least half there was none before it, so the record gives no typical last or first frost date. Midwinter here is the 10-day period half a year from the least frosty point of the year (see the methodology below). It is not always the frostiest 10 days. The cautious dates still apply: they mark the 1-in-10-year late and early frosts, with a 1-week buffer.

Frost-free growing season length by city

The frost-free season is the number of days between the typical last spring frost and the typical first autumn frost (Copernicus C3S frost-day record, 1981-2010).

Leaving out the cities marked †, the typical frost-free season in the grid cells nearest the cities on this page ranges from 112 days in Calgary to 334 days in New Orleans. The rankings are for those grid cells. Station records, compared on this page only for US and Canadian cities, can rank mild cities differently.

The 10 longest typical frost-free seasons in the tables above

The 10 shortest typical frost-free seasons in the tables above

How the grid compares with weather station records

To check the grid against measurements, the table below compares every US and Canadian city on this page with a published 1981-2010 station normal. For a US city, the station is the nearest NOAA first-order weather station (most are airports) within 25 km of the city centre that has freeze normals. The values are NOAA NCEI 1981-2010 Climate Normals: the 50% probability dates of the last spring and first autumn minimum of 32°F (0°C) or lower, and the 50% growing season length. For a Canadian city, the station is the nearest Environment and Climate Change Canada (ECCC) station within 25 km whose frost normals use at least 20 years of record (normals codes A to C). The values are the 1981-2010 Canadian Climate Normals: the average dates of the last spring and first autumn frost (0°C, 32°F), and the average frost-free period.

102 of the 108 cities have a typical last and first frost date in both the grid and the station record. Across them, the median absolute difference is 5 days for the last frost, 6 days for the first frost and 11 days for the frost-free season. The last frost dates agree within 7 days in 66 of the 102. The station’s frost-free season is longer than the grid’s in 54 cities and shorter in 45. A positive difference means the grid date is later than the station date (or the grid season is longer).

23 cities have a typical date more than 14 days from their station, or a typical date where the station records frost too rarely for a median. They are marked † in the tables above, largest gap first: Los Angeles, California (station: frost too rare for a median date); Sacramento, California (last frost +37 days, first frost −13 days); Eugene, Oregon (last frost −34 days, first frost +14 days); Las Vegas, Nevada (last frost +33 days, first frost −16 days); Baltimore, Maryland (last frost +17 days, first frost −30 days); Albuquerque, New Mexico (last frost +29 days, first frost −20 days); Charleston, South Carolina (last frost +27 days, first frost −19 days); Cleveland, Ohio (last frost +17 days, first frost −27 days); San Antonio, Texas (last frost −25 days, first frost +15 days); New York City, New York (last frost +4 days, first frost −23 days); Tallahassee, Florida (last frost −22 days, first frost +16 days); Tucson, Arizona (last frost +21 days, first frost −9 days); Pittsburgh, Pennsylvania (last frost +8 days, first frost −20 days); Chicago, Illinois (last frost +8 days, first frost −19 days); Orlando, Florida (last frost +19 days, first frost −4 days); Phoenix, Arizona (last frost +19 days, first frost −19 days); Reno, Nevada (last frost +19 days, first frost −14 days); Fresno, California (last frost −18 days, first frost +18 days); Philadelphia, Pennsylvania (last frost +7 days, first frost −17 days); Columbia, South Carolina (last frost +10 days, first frost −16 days); Halifax, Nova Scotia (last frost +14 days, first frost −15 days); Newark, New Jersey (last frost +2 days, first frost −15 days); Vancouver, British Columbia (last frost −13 days, first frost +15 days).

Nine of these 23 cities are in the western US, from Eugene in Oregon to Tucson in Arizona. Several others have a station in a city centre or on the shore, such as Central Park in New York, Burke Lakefront Airport in Cleveland, the Customs House in Baltimore and Downtown Charleston. In 18 of the 23, the station’s frost-free season is longer than the grid’s. Two differences in definition also matter. The stations count a minimum of exactly 0°C (32°F) as frost, while the grid counts only temperatures below 0°C. ECCC also reports averages, while the grid and NOAA report medians.

How these frost dates were calculated

Dataset. The frost data is the frost days (FD) indicator from the Copernicus Climate Change Service (C3S) dataset “Agroclimatic indicators from 1951 to 2099 derived from climate projections”, version 1.1 (doi:10.24381/cds.dad6e055), published under a CC BY 4.0 licence. Leaftide uses the historical series driven by WFDEI, the WATCH Forcing Data methodology applied to ERA-Interim reanalysis (Weedon et al. 2014), for 1981 to 2010. FD is the number of days in each 10-day period on which the daily minimum air temperature was below 0°C (32°F).

Resolution. The grid has cells of 0.5° latitude by 0.5° longitude. A cell is about 55 km (34 miles) from north to south and narrower from east to west away from the equator. Time is resolved to 10-day periods, so every frost date is tied to the middle of a 10-day period (the 5th, 15th or 25th of a month), or falls between two of them when a percentile lies between two years. A date one day after the middle of a period, such as April 16 or October 26, usually comes from a leap year, where the middle of the same 10-day period falls one day later in the day-of-year count. The Frost Date Finder shows the same dates.

Season split. For each grid cell, the year is split at its least frosty point, the middle of the longest run of 10-day periods with the fewest frost days. The last spring frost of a season is the last 10-day period with any frost before that point. The first autumn frost is the first 10-day period with any frost after it. A year without spring or autumn frost still counts, so the percentiles cover all 30 years, not only the frosty ones. Midwinter is the 10-day period half a year from the least frosty point. A year with no frost before midwinter gets midwinter as its first autumn frost, and a year with no frost after it gets midwinter as its last spring frost.

Percentiles. The typical date is the 50th percentile (median) across 1981-2010. The cautious last frost date is the 90th percentile plus 7 days, and the cautious first frost date is the 10th percentile minus 7 days. The 90th percentile is the date by which the last frost had come in about 9 years in 10.

Frost-free cities. A grid cell is treated as frost-free when it had frost in at most 3 of the 31 winters in the record (fewer than 1 in 10), or when the 1-in-10-year frost window is empty. The first and last winters in 1981-2010 are partial, which is why the record spans 31 winters. Those cities are listed with the number of winters that had any frost, and no date.

Cities. City coordinates come from the GeoNames cities5000 database. Each city uses the grid cell that Leaftide’s app and the Frost Date Finder pick for those coordinates: the cell whose centre is nearest, unless the terrain check below picks a milder neighbour. Leaftide keeps the city list and the script that builds these tables, so the numbers can be regenerated.

Terrain and station checks. A grid cell averages everything inside it, so a cell that is mostly mountains has later spring and earlier autumn frosts than a town in its valley. Like Leaftide’s app, a city in a valley next to high ground therefore uses the milder neighbouring cell, unless the city sits on the high ground itself. The check applies when the nearest cell has a cautious frost-free season under 120 days and a cell within 0.75° has one at least 30 days longer; the candidate is the cell within 0.75° with the longest cautious season. Elevation from NOAA’s ETOPO 2022 relief model then decides. If the two cells’ mean elevations differ by 150 m (about 490 feet) or more and the city’s own elevation (the mean of the 0.05° block, about 5 km across, that holds it) is closer to its nearest cell’s mean, the city is on the high ground and keeps the nearest cell. Otherwise it uses the milder cell. Flagstaff and Cheyenne keep their nearest cell this way, and their airport records agree with it within 5 days. Salt Lake City, Bolzano, Inverness and Queenstown use the milder cell and are marked †. The note under the table gives the nearest cell’s dates and, for US cities, compares the two cells with the airport record. A US or Canadian city is also marked † when a typical grid date is more than 14 days from its station normal (see the station comparison above). For every city, the tables show the dates the Frost Date Finder shows. The rankings leave out every city marked †.

Baseline years. These dates describe the 1981-2010 climate. This page does not measure any change since then, so treat them as a baseline. For how frost changed between 1981-2010 and 2011-2025, see how frost dates are changing.

Air frost, not ground frost. The data measures air temperature at 2 metres (about 6.5 feet). The UK Met Office notes that when the ground cools faster than the air, a ground frost can form without an air frost (What is frost and ice?). Low-growing seedlings can therefore be hit on nights the air frost record does not count.

Sources

• Copernicus Climate Change Service (C3S): Agroclimatic indicators from 1951 to 2099 derived from climate projections, version 1.1, frost days (FD), WFDEI historical run 1981-2010. Climate Data Store, doi:10.24381/cds.dad6e055. Licence: CC BY 4.0.
• Weedon, G. P., Balsamo, G., Bellouin, N., Gomes, S., Best, M. J. and Viterbo, P. (2014): “The WFDEI meteorological forcing data set: WATCH Forcing Data methodology applied to ERA-Interim reanalysis data.” Water Resources Research 50, 7505-7514. doi:10.1002/2014WR015638.
• GeoNames: cities5000 extract, downloaded 21 October 2026. geonames.org. Licence: CC BY 4.0.
• NOAA National Centers for Environmental Information: 1981-2010 Climate Normals, annual and seasonal freeze products. ncei.noaa.gov/data/normals-annualseasonal/1981-2010/.
• NOAA National Centers for Environmental Information: ETOPO 2022 Global Relief Model, version 1, 60 arc-second surface elevation. doi:10.25921/fd45-gt74.
• Met Office: “What is frost and ice?” weather.metoffice.gov.uk.
• Environment and Climate Change Canada: Canadian Climate Normals 1981-2010, frost-free period. api.weather.gc.ca climate-normals collection.

Cite this data

Leaftide’s frost date tables and the article text are published under Creative Commons Attribution 4.0 (CC BY 4.0). The underlying frost-day data belongs to the Copernicus Climate Change Service and is also published under CC BY 4.0.

How to cite: Leaftide (2026). “Last and First Frost Dates by City (1981-2010 Data).” Retrieved from https://leaftide.com/learn/last-frost-dates-by-city/. Computed from Copernicus Climate Change Service agroclimatic indicators (frost days, WFDEI 1981-2010), doi:10.24381/cds.dad6e055.

Last updated: 21 October 2026.

To look up any location, including your own garden, use the Frost Date Finder. For what to do with your dates, see the last frost date guide and frost damage temperatures.