The objective of early generation testing is to increase breeding efficiency through early identification of superior heterogeneous populations. Bulk F2 testing has gained adherents among the cotton breeding community with little data to support its efficacy. The goal of this study was to compare bulk F2 performance with historical pedigree selection records to determine the degree of correspondence between the two methods in identifying superior hybrid pima cotton (Gossypium barbadense L.) populations. Thirty hybrid populations, originally created in the USDA-ARS Pima breeding project in 1983 and 1984, were recreated. The F2 populations of 1983 and 1984 crosses were tested for yield and fiber performance in randomized complete block tests at Maricopa and Safford, AZ in 2000 and 2002, respectively. Pedigree selection records of the 30 populations were summed across F2, F3, and F4 generations, and correlations were run between the summed selection records and F2 population performance data. Comparisons also were made between F2 performance of populations and the number of lines each contributed to advanced, replicated testing. Significant correlations were almost entirely lacking between selection records of 1983 crosses and the yield and fiber data of their F2 populations. A greater number of correlations were observed between field and fiber lab selection numbers of 1984 crosses and the yield and fiber performance of the F2 populations. These correlations were not consistent between the Maricopa and Safford locations. Weak correspondences were observed between F2 generation performance of the 30 populations and the number of progeny each contributed to advanced, replicated testing.