Using cross-sectional data for U.S. manufacturing, the author identifies a negative effect of unions on the profits of highly concentrated industries and virtually no effect on the profits of unconcentrated industries. The results for concentrated industries are explained in terms of the monopoly model which predicts a negative effect of unions on profits equivalent to the change in employer's surplus. The monopoly model also suggests that previous profit studies which omitted the union variable were likely to significantly underestimate the relationship between concentration and profits. This prediction is also supported by empirical evidence.