An initial collaborative study of the AOAC use-dilution method (UDM), used for bactericidal disinfectant efficacy testing, demonstrated extreme variability of test results among the 18 laboratories testing identical hospital disinfectants. In an effort to improve the method, 32 changes were made by the UDM Task Force. These changes represented improvements in quality assurance practices and elimination of method variability; however, the basic framework of the method was retained. A second collaborative trial was conducted to determine if the interlaboratory variability of test results could be reduced to an acceptable level using the modified UDM. Twelve of the original 18 laboratories participated in the second study. Each laboratory processed 60 penicylinders (P) for each of the 6 randomly selected, federally registered disinfectants and 3 test organisms (Staphylococcus aureus, Salmonella choleraesuis, Pseudomonas aeruginosa). The number of positive penicylinders (greater than 1 positive P/60 replicates = failure) for the 6 products when P. aeruginosa was used as the challenge organism ranged 1-30, 0-36, 0-15, 0-5, 0-3, and 0-60 for the 3 quaternaries and 3 phenolics, respectively. The results of the variance components analysis for P. aeruginosa and the other 2 organisms showed that the variance components for laboratories were not significantly reduced for any organism in this study. Such interlaboratory variability of results questions the use of the original or the modified UDM for registration purposes.