This paper provides the first documentation of the power and specification of the generalized sign test, which is based on the percentage of positive abnormal returns in an estimation period. In simulations using daily stock return data, the generalized sign test is well specified with both exchange listed and Nasdaq stocks. A rank test is more powerful under ideal conditions. However, the rank test is more sensitive to increases in the length of the event window, to increases in return variance and to thin trading. The generalized sign test is a viable alternative to the rank test under these conditions.