Union Election Wins Decrease Again in 2026
Here is my annual review of U.S. union election wins during the previous federal fiscal year. It’s not looking good folks.
Introduction
In this article I review the recent data on U.S. private sector union election wins to assess the state of labor movement organizing. The National Labor Relations Board (NLRB) releases reports on union elections, both monthly and annually, for each federal fiscal year, which last ended on September 30, 2026. As of this writing, the FY2026 annual report still hasn’t been posted, but all the monthly reports are there, so this analysis is based on a compilation of those.
This data doesn’t describe all the new union organizing that’s happening. It excludes the public sector, any transportation sector campaigns covered by the Railway Labor Act, and all organizing that happens through card check or other voluntary recognition — all data that is unfortunately more difficult to gather. But NLRB elections are the main way that private sector organizing happens, so it’s worth examining this data in detail as a good barometer to discern labor movement organizing trends.
My summary from last year is here, which covered the beginning of the second Trump administration. See the Methodology section at the end for more about this data.
Elections and Wins Decrease Again
The chart below shows the number of elections (RC and RM types), where the case is closed and a final result is determined, and the union win rate for every year since 2001.
We can clearly see that over the past few decades there has been a long term trend toward fewer elections and a higher win rate. I have interpreted that trend as unions concentrating their resources on the elections they were more confident of winning and doing much better over time at winning them. There has been more recently a notable increase in elections and win rate from 2022–2024, in what looked like the beginning of a possible new organizing upsurge. This coincided with the end of the Covid-19 pandemic during the labor friendly Biden administration. I have previously discussed these factors that year. In fact 2024 was a modern era high water mark, with the most elections in nearly 20 years and the highest win rate (77%) in many decades. The union win rate over the past year dropped slightly from 76.7% to a still very good 74.3%, and I’m not particularly concerned about that.
But that organizing upsurge has certainly stalled over the last two years, with the number of elections declining in 2025 and again in 2026, dropping about 14% from 1,406 to 1,214. What mitigates this somewhat is taking into consideration the impact of the 43 day Oct-Nov 2025 U.S. federal government shutdown, which stopped union elections for nearly a month and half, or 12% of the year, about the same amount as the elections decrease. The elections that were already scheduled for that time period were able to happen when the shutdown ended. The number of elections for December 2025 and January 2026 were over 20% higher than the remaining months of the year, indicating that those delayed elections were happening. But all the new election petition filings and scheduling that would have happened during the shutdown were delayed until afterwards.
Possibly all those elections were just delayed but still ocurred, but the exceptions would be if any election petitions were pulled (when the union thinks it will lose), as employers had more time during the delay for union busting. It’s also possible that these delays ended up pushing back the scheduling for some election filings later in the year, possibly until the next year. A fuller understanding would require looking at all the election filings, seeing if more were pulled, and seeing if election times were more delayed, but it’s probably not worth it.
So given all this, I think we can conclude that the number of elections would have been fairly similar to the year before without the shutdown, and I may be over-analyzing all this anyway. It is tragic however, if the shutdown resulted in any workers not winning their union that would have, either through canceled elections, or elections lost due to more union busting. There are real possible consequences there, and let’s remind any remaining Trump fans.
Newly Organized Workers Decrease
Get Eric Dirnbach’s stories in your inbox
Join Medium for free to get updates from this writer.
This next chart shows the number of workers eligible to vote in these elections and the number that won their elections.
Again we see the long term decline in both sets of data and also the recent upsurge from 2022–2024. In 2024 the number of workers that won their election was about 94,000, the highest in several decades. But we see a drop in 2025 and then 2026, with the number of eligible workers declining last year 28%, and the number of winning workers dropping 34% to about 49,000. As before, part of this can be attributed to the federal government shutdown. The larger decrease in winning workers can also be explained by the slightly lower win rate. Another reason for the decrease in both numbers is that the average election bargaining unit size dropped from the year before, but that number tends to jump around from year to year, and I’m not sure that’s significant.
Looking at RM Elections
It’s worth mentioning specifically the impact of RM elections, which result from petitions filed by employers when unions make a demand for recognition. There have traditionally been very few in recent decades, about 16 per year during the 10 years before 2024. Then the NLRB’s Cemex decision in Aug 2023 led to a jump in RM petitions as the Board instituted a more labor friendly procedure for voluntary union recognition (see Methodology for more information). Unions made more recognition demands and RM petitions then jumped to 143 in 2024. But they declined to 90 in 2025 and dropped again this last year to 67.
I interpret this as the Cemex framework being less helpful for unions than anticipated, especially with the current Trump NLRB. As I said last year, it’s also possible that employers have granted more recognitions and thus are filing fewer RM petitions. That would be great news, but that data is unavailable in these reports, and I haven’t yet seen any evidence that voluntary recognitions have dramatically increased over the past year. So unions are likely returning to lower levels of recognition demands, at least during this Trump era.
What Does All This Mean?
This recent drop in annual union elections is disappointing but perhaps not a particular cause for alarm, as it’s largely due to the 2025 U.S. federal government shutdown. The win rate is also still very good. The biggest concern is the incredibly small number of workers organized, less than 50,000.
As I have been saying for years, this low level of organizing is typical, far too low, and has not been enough to grow the labor movement for decades. We are trapped in a labor relations system where the principal avenue for union organizing produces too little results. Given the large size of the U.S. workforce (about 130 million non-union workers), we need to organize a million workers per year to start to increase the union membership rate again, which is currently at a low 10% of the workforce. Thus we need well more than 10 times the number of elections and workers organized. That’s a level that looks impossible at this point without a major change in labor movement strategy and a significant increase in organizing resources.
Moreover, as I said last year, Trump’s regime of chaos and cruelty thus far has posed major challenges for unions that have had to undertake significant defensive work and other activities aside from traditional organizing. While the current NLRB is functional and is processing elections, though understaffed as always, we can expect no labor friendly rulings that will help.
Labor strategist Chris Brooks wrote recently about the possibility of the shrinking labor movement crossing an “event horizon” in the future, where it becomes too small to ever recover.
Unions could become too small to win big gains for current members, too weak to organize new workers at scale, too invisible to the public to matter. Beyond this threshold, there may be no climbing back. Not in a generation. Maybe not ever. Our choices in the coming years will determine whether labor has a future, and whether we become the last generations to remember what it meant for the working class to have a fighting chance.
I agree we are in a long running crisis with no sign of improvement. Since 1980 the union membership rate has risen only four times from one year to the next, and then only slightly. We need a massive new organizing program to reverse the labor movement decline, and it will likely include strategies different than running NLRB elections.
More union data on membership and strikes will be released early next year by the U.S. Bureau of Labor Statistics, and I plan to follow up with a further “State of the U.S. Unions” analysis and a longer discussion.
Methodology
I gathered data on union elections from the NLRB Election Reports from 2001–2026, which present the election cases that have closed and have a final result. The NLRB has changed how they present the data over time. For 2001–2010, I used the numbers from the September report (last month of the fiscal year) in the first chart, “FISCAL YEAR 20XX SUMMARY OF ELECTIONS.” I took the total elections, union wins, total employees, and employees won by unions. This data covers both RC and RM elections together. For 2011–2024, from the annual reports I counted the number of RC and RM Elections, election wins, total employees eligible, and total workers that won their elections. For 2025 and 2026 I combined all the monthly reports to get this data, as the annual report was unavailable as of this writing.
Note that the number of workers that won their elections is different from the reported number of Total Valid Votes For, which reports the yes votes in all the elections. It’s important to look at both RC elections (where the union filed the election petition) and RM elections (where the employer filed the petition). After the 2023 Cemex decision, unions can demand recognition from management, which then either has to grant it or file for an election; they can’t ignore it as before. Thus the number of RM elections, which used to be very small, is now much larger, but seems to be declining again. I should also note that the annual closed cases in this data will be different from the number of election petitions filed or the number of elections held each year.
A full analysis of workers gained through these elections would also subtract workers lost in RD decertification elections, but I ignore that here, as that number is fairly small, and I’m interested in the efforts that unions put into new organizing. It’s also possible that some of the election wins for multi-union elections involve already organized workers that were subject to a raid by another union, but I ignore that here, as the number of multi-union elections is very small.
