Core vs Headline Inflation: Why the Two Numbers Diverge (September 2026)

The two numbers on your screen, answered first

In September 2026, U.S. consumer prices rose 3.7% from a year earlier. Core prices, which leave out food and energy, rose 2.6%. Believe both numbers — but use them for different jobs. The 3.7% is what American households actually paid. The 2.6% is the best read of where inflation is headed. Almost the entire gap between them is one thing: gasoline, up 34.2% from a year earlier.

All figures in this post come from the Bureau of Labor Statistics (BLS) Consumer Price Index release for September 2026 (USDL-26-1574, published October 14, 2026), unless noted otherwise.

What each number actually measures

Headline inflation is the change in the full Consumer Price Index for All Urban Consumers (CPI-U): the whole basket — groceries, gasoline, rent, haircuts, car insurance, all of it, weighted by how much households actually spend on each category.

Core inflation is the same basket with two categories removed: food and energy. Its official name is "all items less food and energy." That removal is smaller than it sounds. Using the relative-importance weights in the September 2026 release, food is about 13.5% of the basket and energy about 7.4%, so core still covers roughly four-fifths of what a typical household buys. Core is not a niche subset. It is most of your spending.

The BLS publishes both numbers in the same release on the same morning. Headline remains the official all-items number — the one in the news.

Why economists throw out exactly these two categories

It is not because economists think food and gasoline do not count. It is because those prices move for reasons that often have little to do with the rest of the economy, and almost nothing to do with monetary policy.

Gasoline prices swing with world oil markets, refinery outages, and hurricanes. Food prices swing with droughts, crop disease, and animal illness — egg prices are the textbook example. A central bank raising or cutting interest rates cannot do much about any of those in the short run. So when economists want to see whether price pressure is spreading through the parts of the economy that policy can influence — rent, insurance, car repairs, haircuts — they look past the bumps. That is what core is: the "past the bumps" number.

Two important qualifications most explainers skip:

First, the "food and energy" label was misleading this September. Food prices were flat from August to September. The gap between headline and core came entirely from energy. Gasoline alone rose 9.1% in September and, according to a component-by-component analysis, added about 1.09 percentage points to the 3.7% headline — nearly 30% of the total increase, from a category that is about 3% of the basket. When a 3% slice of your budget adds a third of a point to the inflation rate, the monthly number tells you more about gasoline than about inflation. That is precisely when core becomes the more useful read.

Second, core does not filter out energy completely. Only the raw energy categories (gasoline, electricity, natural gas, heating oil) are removed. The costs they create stay in core. Airline fares were up 22.4% from a year earlier in September — and they are counted in core, not headline-excluded energy. That is fuel cost being passed through to customers, visible in the "underlying trend" number.

The September 2026 gap, line by line

Here is how the two numbers diverged, from the BLS release:

Headline CPI: 3.7% over 12 months, up from 3.4% in August. Month to month, prices rose 0.6% on a seasonally adjusted basis.

Core CPI: 2.6% over 12 months, up from 2.4% in August. Month to month, core rose 0.3% — its second straight month at that pace. Over the three months to September, core prices ran at a 3.4% annualized pace, which is above where it stood earlier in the year.

Inside core, the two sides of the basket behaved differently. Shelter rose 0.4% in the month, up from 0.3% in August, and stood 3.2% higher than a year earlier. Core goods rose 0.3% in the month after 0.1% in August, and just 0.7% over 12 months. The firming came from services — the stickier part of inflation.

For context on the arc of 2026: headline inflation ran at 2.4% in January, jumped to 4.2% in May as the energy shock hit, fell back to 3.4% by August, then rose again to 3.7% in September. Core never rode that rollercoaster — it peaked at 2.9% in May and was back to 2.4% by August. The year is a clean demonstration of why the two numbers are read together.

Which number should you watch?

It depends on what you are using it for. Most explainers pick one winner. The honest answer is that each number answers a different question:

Your paycheck's purchasing power: watch headline. The BLS reported that inflation-adjusted hourly earnings fell 0.5% in September — pay rose 0.1% before inflation while prices climbed 0.6%. That is a headline-number story. Nobody eats core.

Your Social Security raise: watch headline, specifically CPI-W. The 3.6% cost-of-living adjustment announced for 2027 was calculated from the headline Consumer Price Index for Urban Wage Earners (CPI-W) in the third quarter — not core. The gasoline spike that pushed headline to 3.7% helped set the raise that retirees actually receive. For a primer on how the CPI basket itself works, see CPI Explained: What Goes Into the Inflation Basket.

The trend behind the noise: watch core. If you want to know whether inflation is settling or spreading, the 0.3% monthly core pace and the 3.4% annualized three-month pace matter more than one month's gasoline swing. The Federal Reserve watches the trend in core measures for the same reason — and it defines its 2% target in terms of total PCE inflation, a different index, not CPI. In September, the Fed lifted its policy rate to 3.75%–4.00% for the first time since 2023.

What you pay at the pump and the grocery store: watch neither — watch the components. Two broad averages will never tell you what your own bills did. For that, the BLS average-price tables and weekly trackers are more useful — last week's tracker put gasoline at $4.269 a gallon nationally, 30-year mortgages at 7.24%, and eggs at $2.27 a dozen.

What to read in the next CPI report

The October CPI report is scheduled for Tuesday, November 10, 2026, at 8:30 a.m. Eastern. Before reading the headline, read these three lines:

The monthly core rate. If it stays at 0.3% or higher for a third straight month, underlying inflation is firming regardless of what gasoline does.

Gasoline's monthly change. September's 9.1% jump drove more than half of the 0.6% headline increase. If it cools, headline could fall even while core stays put.

Shelter. At roughly a third of the basket, it is the single largest piece of core. It is the slowest to move — and the hardest to bring down once it does.

Three lines, read before the headline: that is the whole technique. Everything else in the release is commentary on those.

Sources

• U.S. Bureau of Labor Statistics, Consumer Price Index — September 2026 (released October 14, 2026): bls.gov/cpi

• Data 4 The People, "What's driving CPI? Every item, every month since 2016" (gasoline contribution analysis): data4thepeople.com

• Portfolio Terminal, "CPI September 2026: 3.7%, Core 2.6%, Gasoline +34%" (component breakdown): portfolio-terminal.com

• US Inflation Calculator, "US CPI Sept 2026: Inflation Hits 3.7% as Gas Prices Surge": usinflationcalculator.com