The simple version

Core CPI leaves food and energy out of the Consumer Price Index (CPI) every month, because those two swing hardest. What it leaves out is decided in advance and never changes.

The Federal Reserve Bank of Cleveland publishes a rival measure. It lines up every category by how far it moved that month and keeps the one sitting in the middle. What falls out is whatever happened to move most, so that list differs every month.

The numbers

• Median CPI rose 0.2% in August 2026, and the 16 percent trimmed-mean CPI also rose 0.2% (Federal Reserve Bank of Cleveland, Median CPI)
• Over the 12 months through August 2026: CPI 3.4%, CPI less food and energy 2.4%, median CPI 2.6%, and 16 percent trimmed-mean CPI 2.6% (Cleveland Fed)
• In June 2026 the one-month figures split four ways: CPI fell 0.4%, CPI less food and energy was flat at 0.0%, the trimmed mean was flat at 0.0%, and median CPI rose 0.2% (Cleveland Fed)
• Median CPI is the one-month inflation rate of the component whose expenditure weight sits at the 50th percentile of price changes (Cleveland Fed)
• The 16 percent trimmed-mean CPI is a weighted average of components whose expenditure weights fall below the 92nd percentile and above the 8th percentile of price changes (Cleveland Fed)
• Both are calculated from the BLS monthly CPI report, and the next update is October 14, 2026 (Cleveland Fed)
• The Cleveland Fed states that by omitting outliers these measures can give a better signal of the underlying trend than either the all-items CPI or core CPI (Cleveland Fed)

Deciding in advance versus deciding each month

Core CPI makes one judgment once: food and energy are the volatile parts, so hold them out permanently. It is simple, repeatable, and comes from the BLS itself.

The cost of that choice is that it cannot see anything else coming. If used cars or airfares or hotel rooms swing hard in a given month, core CPI keeps every bit of that swing, because those categories are not on the list.

The Cleveland Fed version re-decides every month. It ranks the categories by how far each moved, weighted by how much households actually spend on each, and takes the middle one. Whatever swung hardest, in either direction, drops out on its own.

The trimmed mean is the same idea with a wider keep. Instead of one component in the middle, it averages everything between the 8th and 92nd percentiles, discarding 8 percent of spending weight at each end. That is where the 16 percent in its name comes from.

The actual math on one confusing month

June 2026 shows what the difference buys. All four figures below are the Cleveland Fed's; the comparisons between them are ours.

• All-items CPI: down 0.4% on the month.
• CPI less food and energy: flat, 0.0%.
• 16 percent trimmed-mean CPI: flat, 0.0%.
• Median CPI: up 0.2%.
• The spread between the lowest and highest of the four is 0.6 percentage points, for one month in one economy.

Someone told prices fell in June and someone told the typical category rose both heard an accurate statement. The first covers the whole basket including its biggest movers. The second covers the category in the middle, after those movers drop out.

What this means

When a monthly inflation figure lands, more than one defensible number exists for the same month, and they can point different directions. Which one gets quoted usually depends on who is quoting it rather than on which is more correct.

The useful question is what a measure throws away and how it decided. A fixed exclusion list is predictable and can miss the month's real story. A rule that re-decides each month catches surprises but changes what it is measuring as it goes.

What this is NOT

This article explains how median CPI and the 16 percent trimmed-mean CPI are built and how that differs from core CPI. It does not predict inflation, the October 14 CPI release, Federal Reserve decisions, or interest rates. It takes no position on which measure is better or on how price statistics should be constructed, and the Cleveland Fed's own view that its measures can give a better trend signal is reported as the issuer's stated position rather than endorsed here. It is not advice about any financial decision. The 0.6-point spread and the comparisons between the four June figures are our arithmetic on the Cleveland Fed's published data. The Cleveland Fed notes that its figures for October 2025 use interpolated data because no CPI report was published for that month. This is not a political endorsement or criticism of anyone.

Sources

• Federal Reserve Bank of Cleveland, Median CPI (definitions, calculation method, and monthly data), accessed October 6, 2026: https://www.clevelandfed.org/indicators-and-data/median-cpi
• U.S. Bureau of Labor Statistics, Schedule of Releases for the Consumer Price Index: https://www.bls.gov/schedule/news_release/cpi.htm