How to Read US CPI Inflation Data

Read the series before the inflation rate

US CPI measures the average change in prices for a defined consumer population and basket of goods and services. An index level is not itself an inflation rate. To read a release, identify the series, the period being compared and whether the figures are seasonally adjusted before deciding what a percentage means. [1][2]

Core CPI usually means all items less food and energy. It answers a different coverage question from all-items CPI, rather than providing a universally better number. A lower positive inflation rate still means prices increased over the stated period; it does not mean the average price level fell.

This guide uses fictional index levels to explain monthly and yearly changes. It supplies no current inflation reading, consensus forecast or prediction of a market move.

Build a complete series label

Start with the population. CPI for All Urban Consumers, or CPI-U, covers a broad urban consumer population. CPI for Urban Wage Earners and Clerical Workers, or CPI-W, covers a subset with specified earnings and employment characteristics. The Chained CPI for All Urban Consumers, C-CPI-U, uses the CPI-U population but a different index construction that accounts for substitution across item categories. These names are not interchangeable. [2]

The reference base tells you which period is assigned an index of 100. It is not the date of the expenditure data used for spending weights. Neither a base label nor the word “basket” means household spending weights remain fixed forever. This reading guide does not specify a current weight-update schedule. [2][3]

CPI describes an average, so it may not match your household's spending mix or experience. An index level also does not rank living costs across cities: a higher index in one area need not mean that area is more expensive than another.

Keep all-items and core coverage visible

In a typical US CPI discussion, “headline” refers to all items and “core” to all items less food and energy. Excluding food and energy does not mean those purchases are unimportant. It changes which price movements the series measures. [2]

Core is not shorthand for removing housing, used vehicles or any other components a reader chooses. Other exclusion series can exist; use their complete names instead of silently substituting one for ordinary core CPI. You must still identify the population, area and comparison period.

Why an oil-price move is not the CPI move

A crude-oil quote is an input price, not the price households pay for gasoline. The CPI motor-fuel index measures consumer fuel prices. Pump prices also include refining costs and profits, distribution and marketing, and taxes. Changes in these components mean a 10% crude-price rise need not produce a 10% gasoline-price rise. [7][8]

All-items CPI covers a much wider basket. A fuel category’s impact depends on its weight and price change; other categories can add to or offset that movement. For actual data, match the category, period and dated weights. [9] To illustrate weighting, assume a fictional fixed-quantity basket initially costs 100 units: 3 for retail fuel and 97 for everything else. If only retail fuel prices rise 10%, its cost becomes 3.3 and the basket costs 100.3, an increase of 0.3%. The 3% starting fuel share is invented, not an actual US weight. This is neither the full BLS index-construction method nor an assumption that a 10% crude move causes that retail increase.

Excluding food and energy from ordinary US core CPI removes those categories, not every possible influence of energy costs on included prices. Energy can enter production and transport costs; whether firms pass those costs to customers depends on their circumstances. Federal Reserve staff research discusses possible cost and broader economic channels, using historical PCE and HICP data rather than a current US CPI coefficient. [10] Neither a fixed delay nor a one-for-one headline or core response follows from the index definitions.

Match the comparison period and adjustment

A month-over-month change compares a month with the immediately preceding month. A year-over-year change compares it with the same month one year earlier. The reference month is the month measured, which is different from the date on which BLS publishes the release. [2][4]

Seasonal adjustment helps distinguish short-term changes from recurring patterns such as those associated with holidays or seasons. SA and NSA are different versions of a series, not labels you can remove once you have found a number. The table shows common presentation choices; the actual release label remains decisive. Match the adjustment at both endpoints of each calculation. [2]

January to December within one year spans 11 monthly intervals. A 12-month comparison uses, for example, December to the following December. An annual-average change instead compares the average index across one calendar year's 12 months with the corresponding average for another year. It need not equal a December-to-December change. [4]

Do not add 12 monthly percentages to calculate a yearly change. Compatible, unrounded monthly index ratios can be multiplied, but adding rounded headlines is a different operation. Mixing SA monthly changes with an NSA annual change will not reconstruct the annual headline either.

Calculate a change between index levels

For any positive starting index, the percentage-change formula is 100 × (ending index ÷ starting index − 1). For a monthly change, the starting observation is the previous month; for a 12-month change, it is the same month one year earlier. Keep the series, coverage, reference base and adjustment consistent. This is a calculation between index values, not the full method BLS uses to construct its price indexes. [4]

Assume a fictional all-items CPI-U, U.S. city average series on one consistent reference base. All three inputs below use the same adjustment basis. This isolates the period arithmetic; it is separate from pairing an actual SA monthly headline with an NSA yearly headline.

Keep full precision until the display step. The example rounds to one decimal place; do not treat that as a rule for every BLS figure. Published index levels and reported changes can have different display precision, so a calculation from rounded inputs may not reproduce every published change exactly. [4]

Separate forecasts, releases and revisions

A forecast or “consensus” is an expectation from a named forecaster or survey, observed at a stated time. It is not a BLS price observation. Before comparing it with a release, match the series, period, adjustment and units. This guide uses no real consensus figure.

In a separate fictional comparison, suppose a matched YoY headline is reported at 3.7% against a forecast of 3.5%. The difference is +0.2 percentage points. It is not a relative increase of 0.2%, and it is not the 0.2% monthly change in the index example. Matching digits do not make the quantities identical.

Revision rules depend on the series. CPI-U and CPI-W NSA data are not routinely revised to incorporate late responses, although errors can be corrected. Seasonally adjusted histories can be revised when seasonal factors are re-estimated. Chained CPI has a separate preliminary-estimate and revision process. “CPI is never revised” is therefore too broad. [2]

Keep the release vintage when comparing two readings. A new value, a revised earlier value and a forecast made before the release answer different questions; record which one you are using.

Why PCE inflation can differ from CPI

PCE and CPI can differ because of their formulas, spending weights, coverage and other statistical effects. Neither is simply the other index with a different name. The PCE price index measures spending by and on behalf of the personal sector, including households and nonprofit institutions serving them; CPI focuses on out-of-pocket purchases by its covered households. [5]

The Fed's 2% longer-run inflation objective uses PCE prices. It is not a 2% CPI or core-CPI target, a cap on every observation or a guaranteed future path. The policy objective and the definition of an index are different things. [6]

For a comparison, match the period, identify the measure and check the release vintage. Do not assume PCE is always lower or more accurate. The useful question is which series answers the question you are asking.

Read the release without assuming a market reaction

Inflation is one part of the information considered in monetary policy and market pricing. A release can change expectations, but “above forecast” is not a rule that fixes the next Fed decision or the direction of stocks, currencies or gold. The forecast benchmark, other information and the outcome being assessed all matter. [6]

For meeting-specific rate interpretation, our FedWatch guide separates futures-derived estimates from policy promises. A statistical release also does not by itself define an event contract. The exact series, reference month, threshold, release vintage and revision treatment would need to match that contract's wording.

Our prediction-market guide explains the qualifications around prices and events. For the task of reading an individual contract's rules, the rule-reading steps provide further context. These links do not establish that a CPI market is offered or that any particular reader can use a service.

Knowing how an inflation figure is calculated does not remove trading risk. Real-money event contracts can lose money and incur fees; they are not a dependable income source or a way to recover earlier losses.

Common questions about US CPI

Is CPI itself an inflation rate?

No. CPI is published as an index; an inflation rate describes its percentage change over a stated period. An index of 300.600 is neither $300.600 nor inflation of 300.600%. You need a matching earlier index to calculate a change.

Why can monthly and yearly inflation differ?

They compare different starting points. The monthly rate uses the previous month, while the yearly rate uses the same month one year earlier. A release may also present SA monthly and NSA yearly figures, so retain the adjustment labels.

Does slower inflation mean prices fell?

Not if the rate is still positive. A lower positive rate means a smaller increase over the stated comparison period. Falling prices would require a negative change in the relevant index over that period.

What is removed from core CPI?

Ordinary core CPI excludes food and energy. It does not remove every volatile category a reader might choose. Check the complete series name, population and period before comparing figures.

Why can PCE differ from CPI?

The formulas, spending weights, coverage and other statistical effects differ. That can produce different readings without either release being a forecast or a measure of every household's own inflation.

Does an above-forecast CPI figure guarantee a market move?

No. A forecast comparison is not a guarantee of a Fed decision or an asset-price direction. In a matched fictional comparison, 3.7% actual versus 3.5% forecast is +0.2 percentage points, not a trading signal.

Use a short reading checklist

1. Copy the full US series label: population, area and item coverage.
2. Identify the reference month, publication date and release vintage.
3. Read the adjustment and comparison period before the number.
4. Separate index points, percentage changes and percentage-point differences.
5. Check whether a comparison uses a forecast, an earlier release or a revised value.
6. Keep any policy or contract interpretation tied to its own assumptions and wording.

That sequence is more useful than reacting to an unlabeled headline. It lets you explain what changed, what was compared and what the number cannot establish.

Sources and scope

Sources checked on 18 October 2026. BLS documentation supports the CPI definitions, period comparisons and index arithmetic; BEA supports the PCE comparison, and Federal Reserve material supports the policy-objective distinction. All worked index levels, inflation rates and forecast comparisons here are fictional. The BLS arithmetic tutorial and BEA reconciliation contain historical examples; they are not evidence of current weights, schedules or numerical gaps. No live release or consensus is reported.

Oil-price and consumer-fuel source check: 20 October 2026. The added fixed-basket example is hypothetical; its 3% fuel share is not an actual US weight. The Federal Reserve staff note supports possible transmission channels using historical PCE/HICP analysis, not a current CPI coefficient. Earlier references retain the source-check date stated above.

1. CPI Home
2. Consumer Price Index Frequently Asked Questions
3. Consumer Price Index: Overview
4. Math calculations to better utilize CPI data
5. What accounts for the differences in the PCE price index and the Consumer Price Index?
6. Monetary Policy: What Are Its Goals? How Does It Work?
7. EIA: Factors affecting gasoline prices
8. BLS: Measuring Price Change in the CPI — Motor fuel
9. BLS: What are contributions and effects?
10. Federal Reserve staff: Oil Price Pass-Through into Core Inflation (2017; historical PCE/HICP study)