Approval of labor unions in the United States has reached 71%, close to what Gallup measured in May 1965. Only 23% of Americans now want unions to have less influence, the lowest share of the new millennium, down from 26% in 2023 and far below the 42% recorded in 2009, near the century's high. Another 47% want unions to have more influence than they do. That combination puts public opinion roughly where it stood six decades ago. Union coverage has not followed: around 60 years ago close to a third of Americans were covered by a union. In 2025 it was one in ten.
The 1965 comparison is the part worth sitting with. That year was not the start of labor's ascent, it was the top. The average duration of worker-organized strikes peaked in 1965 at 25 days, according to the Bureau of Labor Statistics, at the end of a decade in which work stoppages had been gradually increasing. Support for the protection of labor rights was distinctly stronger then, and the long contraction of the organized labor movement was already beginning. After the 1960s, union influence fell sharply, with movement activity and public sympathy declining together.
That density had not come from sentiment in the first place. It was built over decades of organizing and concessions extracted under conditions of expanding industrialization, by ordinary workers who spent their lives on it without any assurance the effort would produce a result.
Which is why the 71% is a weaker number than it looks. A poll captures willingness to answer a question, and the cost of telling a pollster that unions should have more influence is zero. The cost of building one is measured in years of uncertain, unpaid work, which is precisely what the historical record here describes. For most of the past half century approval and membership fell together, which at least made them a coherent pair. They have now come apart: sympathy at a level not seen since 1965, coverage at roughly a third of the 1965 rate. Something with 71% approval and 10% coverage is not a movement with leverage. It is a constituency.
The automation framing is what makes this poll more than trivia. Years of deindustrialization, and now an economic policy environment in which the threat of AI-driven automation is sharpening, have pushed workers into a defensive posture, and once-active union organizations increasingly give way to business-aligned forces. But the numbers describe attitudes and nothing else. There is no figure here for new bargaining units, no measure of whether anyone is acting on the sentiment, no sign that the 47% who want more union influence intend to supply any of it themselves. Approval is the one labor statistic that asks nothing of the person providing it.
The historical precedent cuts against the optimistic read. Approval was at this level in 1965 and the movement it described was already on the way down, which suggests a high-water mark in opinion is a lagging measurement of power accumulated years earlier and about to be spent. If the workers most exposed to automation are going to get protection through organizing at the workplace, the organizing has to be underway before the displacement rather than in response to it, and the nine in ten with no coverage are starting from nothing. The 71% is not the beginning of a revival. It is a count of the people who would like someone else to have already done the work.