Jobless claims: reading the weekly unemployment insurance report

Jobless claims are the number of people filing for state unemployment insurance (UI) benefits. The US Department of Labor publishes them every Thursday at 8:30 a.m. Eastern, for the week that ended the previous Saturday, which makes them the most timely official reading on the labor market. The October 29, 2026 release, for example, reported 194,000 seasonally adjusted initial claims for the week ending October 24.

Because they come from administrative records rather than a survey, and arrive every week rather than every month, claims are the first place a wave of layoffs shows up. They rose months before the official start of the 2001 and 2007–2009 recessions and jumped from about 200,000 to more than 6 million in four weeks in March and April 2020.

This guide explains the two main claims series, the seasonal adjustment and the 4-week average, how to read a release, and what the long record shows.

What the claims report counts

A person who loses a job can file a claim with their state's unemployment insurance program. The weekly release reports several series:

• Initial claims: new claims filed in the week. This is the headline and the best proxy for new layoffs.
• Insured unemployment, also called continued claims or continuing claims: the number of people claiming benefits for a week of unemployment. It is reported with an extra week's lag: the October 29, 2026 release gave 1,701,000 for the week ending October 17.
• The insured unemployment rate: insured unemployment as a share of jobs covered by UI, 1.1% in that release.
• Unadjusted figures, state detail and separate federal programs, such as claims by former federal civilian employees and newly discharged veterans (442 and 387 initial claims in the week ending October 17, 2026).

Claims are not the unemployment rate. The rate comes from a household survey and counts everyone jobless and searching, whether or not they receive benefits; many unemployed people never file, because they quit, are new to the labor force, or are not eligible. The The unemployment rate: U-3, U-6 and labor force participation guide covers that measure.

Seasonal adjustment and the 4-week average

Claims have strong seasonal patterns: layoffs after the holidays, auto plant retooling in summer, school-year endings. The Labor Department publishes seasonally adjusted figures that remove the typical pattern for each week.

Even after adjustment, single weeks jump around with holidays, storms and state processing quirks. The 4-week moving average smooths that out.

Claims since 2000

The chart shows the 4-week average, with three episodes standing out.

• 2001. The average reached its low of 266,250 in April 2000, almost a year before the March 2001 business-cycle peak, then climbed to 489,250 in October 2001.
• 2007–2009. The average bottomed at 286,500 in February 2006, nearly two years before the December 2007 peak, and drifted up through 2007. It accelerated in 2008 and peaked at 659,250 in the week ending March 28, 2009. Weekly claims averaged about 574,000 across 2009.
• 2020. Weekly claims went from 208,000 in the week ending March 7, 2020 to 2,914,000 on March 21, 5,946,000 on March 28 and a record 6,137,000 on April 4. The chart's scale stops at 700,000 so that the rest of the history stays readable. Continued claims peaked at 23.1 million in the week ending May 9, 2020.

Since then claims have run low. Weekly claims averaged about 214,000 in 2022, 223,000 in 2023 and 2024, and 226,000 in 2025, close to the 2019 average of about 218,000. The highest 4-week average was 240,000 in early August 2024 and 242,250 in mid-June 2025. In 2026 the weekly range has been 189,000 (week ending July 18) to 230,000, with the 4-week average peaking at 224,500 in June and easing to 196,750 by late October.

How to read a claims release

1. Look at the 4-week average first. A single week can swing by 20,000 or more for reasons that say nothing about the economy.
2. Check last week's revision. The prior week is routinely revised, usually by a few thousand.
3. Watch holiday and storm weeks. Thanksgiving, year-end and hurricanes regularly distort a week or two; the seasonal factors cannot fully handle moving holidays.
4. Compare with continued claims. Rising initial claims with flat continued claims suggests laid-off workers are being rehired quickly; both rising suggests jobs are getting harder to find.
5. Look for a trend off the low. Historically, the warning sign has been a sustained rise in the 4-week average from its cycle low, not any particular level. In the data above, the average rose from its trough well before the 2001 and 2007 peaks.

Why markets care

Claims are the only official labor data released weekly, so they fill the gap between monthly jobs reports. In the week the payroll survey is taken (the week including the 12th), claims are a direct hint about the coming report. They usually move markets less than payrolls, but a sudden jump can shift expectations for the Fed and move Treasury yields and stock-index futures at 8:30 a.m. The Leading, lagging and coincident indicators: what turns first guide shows how claims compare with other early-warning series.

Pitfalls and limitations

• Eligibility, not unemployment. Claims include only people who file and qualify under state rules. People who quit, are new entrants, are self-employed or have exhausted benefits do not appear.
• State rules and processing. Changes in state systems, filing backlogs and fraud waves can move claims without any change in layoffs.
• Seasonal factors. Adjustment is estimated, and unusual years such as 2020 can distort factors for years afterward.
• Low claims can coexist with weak hiring. As 2025–2026 shows, claims tell you about separations, not about how many new jobs are created.
• Continued claims understate long unemployment. Once someone exhausts regular state benefits, whose length varies by state, they drop out of the count even if still jobless.

Claims compared with other labor data

How Finabase does it

Finabase does not publish claims or other economic data, and has no economic calendar. The chart on this page is ours, drawn from the public Labor Department series on FRED as downloaded on October 30, 2026.

What the app does show is how markets trade at the 8:30 a.m. release. The futures page lists stock-index futures such as the S&P 500 and Nasdaq contracts, which are trading before the stock market opens, and the Treasury page shows yields from 1 month to 30 years. For coverage of the release, see the economy topic in Discover.

Frequently asked questions

When are jobless claims released?

Every Thursday at 8:30 a.m. Eastern Time by the US Department of Labor, covering initial claims for the week ending the previous Saturday. Continued claims in the same release cover the week before that.

What is a normal level for initial claims?

There is no fixed normal. In the calmer years since 2018 (excluding 2020 and 2021), weekly claims have stayed between about 190,000 and 260,000; the 2019 average was about 218,000. In the 2000s levels ran higher, averaging about 299,000 in 2000 and 321,000 in 2007. The 4-week average was 196,750 in late October 2026.

What is the difference between initial and continuing claims?

Initial claims are new applications for unemployment benefits filed during the week, a proxy for layoffs. Continuing claims count people still receiving benefits, a measure of how long laid-off workers stay unemployed.

Do jobless claims predict recessions?

They often rise before recessions, but not on a fixed schedule. In our data the 4-week average bottomed about 11 months before the 2001 recession began and about 22 months before the 2007 one, then rose. Their weekly noise means a few bad weeks are not a signal on their own.

Summary

Weekly jobless claims count new filings for state unemployment benefits and are released every Thursday at 8:30 a.m. Eastern, which makes them the fastest official signal of layoffs. The 4-week average smooths weekly noise: it was 196,750 in late October 2026, against 659,250 at the 2009 peak, while weekly claims hit a record 6,137,000 in April 2020. Claims measure firing, not hiring, and only for people who file, so read them with continued claims, payrolls and the unemployment rate rather than on their own.