In American elections, labor union members are more likely to vote for Democratic candidates, and Democrats tend to represent states and districts with more union members. But does the strength of organized labor have a causal effect on election results? I use union certification regression discontinuities to measure union effects and find that private-sector unions do have a positive effect on Democratic vote share in presidential races: an additional union increases Democratic vote share at the county level by 1.5 percentage points. This suggests that each new union member ``converts'' 14 new Democratic voters with the conversion rate decreasing in union size. The effects on Congressional elections are not significant. I find that an additional union increases contributions from labor PACs to Democratic congressional candidates, with no effect on Republican candidates. While there is no union effect on overall partisan position of congressional representatives, I do find that unions push Democrats to the left on labor, trade, and general economic issues.