An approach to determine the appropriate sample sizes for a series of screening trials to identify promising new therapeutic agents was presented by Yao, Begg, and Livingston (1996, Biometrics 52, 992-1001). This approach is now improved to a two-stage design that further minimizes the time to identify a promising agent under fixed error rates. When applied to data from the historical experience of exploratory vaccination trials at Memorial Sloan-Kettering Cancer Center, the method demonstrates that relatively small individual screening trials are optimal. The reliability of the results is evaluated using the bootstrap.