This article examines how bicycle commuting is associated with bike lane access and sociodemographic advantage at the block group level in twenty-two U.S. cities. Using regression models with interaction terms, I find that associations between bike lanes and bicycle commuting are stronger among more advantaged block groups, due primarily to limited bicycle commuting in disadvantaged block groups even when bike lanes are present. This indicates the importance of considering heterogeneous effects in planning research and suggests that bike lane investment, in isolation, could widen sociodemographic disparities in cycling and its benefits if non-infrastructure barriers to cycling are not also addressed.