Nonfarm Payrolls: The Monthly US Jobs Report
Nonfarm payrolls is the US Bureau of Labor Statistics' monthly estimate of how many jobs employers added or cut, published in the Employment Situation report. It usually comes out on the first Friday of the month at 8:30 a.m. Eastern, and the first figure is revised in each of the next two reports.
Nonfarm payrolls is the monthly change in the number of jobs on US employer payrolls, outside farming, as estimated by the Bureau of Labor Statistics. It is the headline figure of the Employment Situation report, the release most people mean when they say “the jobs report”.
Markets watch it because jobs feed spending, and spending feeds inflation and the path of interest rates. A single number can therefore move stocks, bonds and the dollar within seconds of 8:30 a.m. Eastern.
How it works
The report combines two surveys. In the release’s own words, the household survey measures labor force status by demographic characteristics, and the establishment survey measures nonfarm employment, hours and earnings by industry.
Payrolls come from the establishment survey. Employers report how many people they paid, and the change from the previous month is the headline: “+162,000” means the count rose by that many jobs.
The unemployment rate comes from the household survey. That is why the two can point in different directions in the same month, since they are measured from different samples.
The number is seasonally adjusted. Some hiring and layoffs repeat on the calendar every year, and the adjustment strips out those routine swings, so the headline shows the change beyond what the time of year alone would bring.
The level is large, and the changes are small against it. Total nonfarm employment stood at about 159.1 million in August 2026, up from about 109.2 million in January 1990. A monthly change of 100,000 is less than a tenth of one percent of that total.
The release calendar
It is almost always a Friday morning. Of the 391 releases on the BLS archive page, from 4 February 1994 to 4 September 2026, 376 came out on a Friday, 12 on a Thursday, 2 on a Tuesday and 1 on a Wednesday. Nearly all were released at 8:30 a.m. Eastern, an hour before US stocks open their regular session. The exception in the archive is the October 1998 report, issued at 1:30 p.m. on Thursday 5 November 1998.
Holidays move it. Ten of the 12 Thursday releases came in early July, in years when that Friday was the Independence Day holiday; the other two were that 1998 report and the delayed September 2025 report. Ten releases fell on Good Friday, when the US stock market was closed, so SPY’s first regular session after the report was the following Monday.
Government shutdowns delay it. The September 2013 report came out on Tuesday 22 October 2013. In 2025 the September report came out on Thursday 20 November, BLS did not publish a separate October release, and household survey data for October 2025 were not collected, according to the November report of 16 December 2025.
The next dates on the BLS schedule, read on 26 September 2026: the September 2026 report on Friday 2 October 2026, October’s on 6 November and November’s on 4 December, each at 8:30 a.m.
Revisions: the first number is a draft
Each report revises the two months before it. Late survey replies arrive and seasonal factors are recalculated, so the first estimate for a month is updated in each of the next two releases. Once a year the series is also re-anchored to nearly complete employment counts from unemployment insurance tax records, a benchmark revision that can move months that were already revised twice.
2026 shows how far a draft can move. June 2026 was first reported at +57,000 on 2 July, cut to +20,000 on 7 August, then raised to +31,000 on 4 September. July was first reported as a loss of 23,000 and a month later became a gain of 21,000.
So a “miss” at 8:30 a.m. can later vanish. The market reacts to the first print, but the history books keep the revised one.
A worked example
The August 2026 report, released Friday 4 September 2026. The BLS release opened with the line that total nonfarm payroll employment increased by 162,000 in August, with the unemployment rate unchanged at 4.1%.
The same release revised the two months before it. June went from +20,000 to +31,000 (+11,000) and July from -23,000 to +21,000 (+44,000), so employment in June and July combined was 55,000 higher than previously reported.
The arithmetic from the levels. The BLS series put total nonfarm employment at 158,913,000 in July and 159,075,000 in August. The difference is 159,075,000 - 158,913,000 = 162,000, the headline.
How SPY opened. SPY had closed at $773.17 on Thursday 3 September. On the release morning it opened at $772.01, a gap of -0.15% ($772.01 / $773.17 - 1), and closed at $770.19, down 0.39% on the day. The largest monthly gain since March did not lift the index that day, which is the usual lesson: the reaction depends on what was expected, not on the number alone.
The original data
Monthly changes since 1990. From the BLS series, 439 monthly changes from February 1990 to August 2026 (values as revised by 26 September 2026). The median month added 162,000 jobs, and 85 months (19.4%) lost jobs. The largest loss was April 2020 at -20,469,000 and the largest gain June 2020 at +4,631,000. By decade the median month added 211,000 in the 1990s, 41,500 in the 2000s, 196,000 in the 2010s and 210,000 from 2020 to August 2026. All of 2025 added 116,000 jobs in total; the first eight months of 2026 added 643,000. The full series is in the monthly payrolls file.
SPY on release days. Every release date on the BLS archive page was matched to SPY daily bars, 4 February 1994 to 25 September 2026. The 10 Good Friday releases have no session, leaving 381 release days. Each day’s opening gap is the open against the prior close, and the move is close against the prior close. The 1:30 p.m. release of 5 November 1998 came after the open, so its gap does not reflect the report; it is kept in the file. Every day is listed in the release-day file.
- Median opening gap: 0.42% on release days, against 0.29% on the 1,280 other Fridays and 0.28% on all 7,833 other days.
- Gap of 0.5% or more: 42.0% of release days (160), against 29.3% of other Fridays and 27.2% of all other days.
- Close-to-close move of 1% or more: 36.5% of release days (139), against 26.0% of other Fridays and 26.6% of all other days.
What it shows. Release days are measurably more active, mostly at the open, which is when the news is priced. Most release days were still quiet: the median one moved 0.68% close to close.
Little video covers it. In the 24,971-video study behind this site, 3 titles mention nonfarm payrolls, NFP or the jobs report, at a median of 33 views.
When it fails
The first print gets revised away
The September 2025 figure was first reported at +119,000 and now stands at +76,000. January 2026 was first reported at +130,000 and now stands at +160,000. A trade built on the first print is built on a number that often does not survive.
The day holds other news
A close-to-close change records everything that happened that day, not only the report. The largest release-day move in the file, -5.85% on 4 April 2025, shows the limit: a day-level test cannot separate the jobs number from anything else that broke during the session.
The reaction runs against the number
The August 2026 report showed the largest monthly gain since March and SPY still fell. Markets trade the gap between the number and what was expected, and the expectation is not in this data. The guide to trading the news covers why the first move is so hard to capture.
The schedule changes
Shutdowns, holidays and reissued releases all moved the date in the archive. Check the BLS schedule each month rather than assuming the first Friday.
Related
A recession is the broad downturn that falling payrolls are one part of, and that page covers how the NBER dates one.
The consumer price index is the other monthly release that moves rate expectations, and it measures the inflation side of the same question.
Interest rates are the channel through which a jobs surprise reaches stock and bond prices.
An opening gap is what a release-day reaction looks like on a chart, because the report lands before the regular session opens.
Check the revisions to the two prior months before reacting to the headline number. If you hold a position into 8:30 a.m. on release day, size it for an opening gap you cannot exit during, because the report lands while the regular session is closed.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.