Nine point nine percent of US wage and salary workers belonged to a union in 2024, about 14.3 million people, down from 10.0 percent the year before, according to the Bureau of Labor Statistics' annual union members release (BLS, published Jan. 2025). The headline decline sits oddly beside the organizing wave of the same years, and the tension between the two facts explains most of what matters about the numbers.
Where is union membership concentrated?
In government, overwhelmingly. Roughly one in three public-sector workers is a union member, against about one in sixteen in private industry, a ratio the BLS has recorded in similar form for decades (BLS, Union Members 2024). Education, protective services and utilities lead the occupational list; food service and retail sit at the bottom despite generating the most headlines. The geography follows the same pattern, with membership rates in Northern and Pacific states far above the Southern states where most recent plant organizing has happened.
Why did the rate fall while organizing surged?
Arithmetic, mostly. The board reported unions winning about nine in ten contested elections in recent fiscal years (NLRB data, reported by Reuters, July 2024), but new certified units are small relative to the workforce, and union jobs disappeared faster than new ones formed. A cafe or store unit of dozens barely registers against sector-level attrition in manufacturing and government. Membership is a stock; organizing is a flow, and in 2024 the flow did not offset the stock's erosion.
What does the annual release not capture?
Three things, each consequential. Workers in freshly certified units who have not yet been organized into paying membership do not always appear as members. The survey excludes most gig and misclassified workers, precisely where some of the loudest campaigns have run. And the release measures membership, not contract coverage, so a certified-but-contractless workforce can look the same as an unorganized one in the table.
Which figures do the parties each emphasize?
Union-side analysts cite the election win rates and petition volumes at the labor board as the leading indicator; employer-side commentators cite the falling headline rate as the lagging reality. Both readings come from government data, and neither is wrong. The honest summary is that representation activity rose while covered membership eased, and 2026 will show whether the flow finally bends the stock (BLS, Union Members 2024; NLRB election data, 2025).
What the next release will decide
The 2025 edition of the BLS release, published in late January 2026, shows whether the 2024 dip was noise or trend, and whether first contracts signed in the Starbucks and Amazon campaigns converted certified workers into members at scale. Until then, the established facts are these: 9.9 percent, 14.3 million members, one in three government workers organized, and a private sector where organizing energy has outrun membership arithmetic for three straight years.
Which sectors moved the 2024 total?
Government and transport carried the largest member counts, and manufacturing continued its long erosion, on the BLS tables (BLS, Union Members 2024). The fastest-growing organizing targets, coffee shops, retail, logistics and technology-adjacent warehouses, hold workforces that are young, dispersed and expensive to organize unit by unit. A thousand-person hospital campaign moves the stock; a hundred coffee shops of thirty workers each, even all won, barely dents it.
Does the release measure membership or contract coverage?
Membership, defined as workers who answer that they belong to a union, which is not identical to workers covered by a collective bargaining agreement. The covered figure runs slightly higher in most years because it includes workers represented by a union who have not joined, and it includes contract-covered workers in right-to-work states. Analysts who report one number and call it the other produce a small but recurring distortion in coverage of the beat.
Why do the same numbers feed both optimism and pessimism?
Because the release and the board's data measure different things on different clocks. Election petitions and win rates at the NLRB (NLRB, 2025) move quarterly and responded sharply to the post-2021 organizing wave; the BLS rate moves annually and reflects retirements, hiring and sector shifts no campaign controls. A fair reading holds both: the surge is real in the flow data, and so is the arithmetic drag in the stock data.
What would change the headline rate?
Scale, not victories. The rate moves when unions organize units in the tens of thousands, or win coverage in fast-growing sectors, or lose it in declining ones. Two Amazon elections and several hundred Starbucks stores did not appear in the 2024 rate as movement, and the release says nothing about whether they will in 2025. That question is for the next table, not this one.
What do the earnings tables add to the picture?
The annual release pairs membership with median weekly earnings for full-time wage and salary workers, and the union figure runs above the nonunion figure in every recent edition (BLS, Union Members 2024). Analysts caution against reading the gap as a union premium, since union density concentrates in sectors, ages and public-sector jobs that pay more for reasons the table does not isolate. The table is best used as a description of who is organized, not a measure of what organizing pays.
The same caution applies to the sector tables. Government's high membership rate reflects legal frameworks that ease recognition as much as worker preference, and the private sector's low rate reflects the reverse. The release describes the workforce as it is; the why belongs to statute and strategy.
Related: How a union representation election actually works · Can public employees be required to pay union fees?.
For more context, read How a union representation election actually works.
For more context, read Can public employees be required to pay union fees?.
For more context, read How to read a union's LM-2 financial filing.