The presence of detectable HIV-1 prior to the appearance of HIV-1-specific antibody was assessed in 41 incident infections that occurred during a 6-year prospective cohort study. All available antibody-negative samples (n = 138) and the first antibody positive sample (n = 41) were tested, under code, by the polymerase chain reaction (PCR) in two laboratories and by HIV-1 isolation in a third laboratory. Samples were available as long as 66 months and at least 18 months before seroconversion for 24/41 subjects. An equal number of time-matched control specimens from persistently seronegative homosexual men and 103 samples from normal blood donors were also tested under code. Samples with discordant results were subjected to coded repeat analysis along with appropriate controls. All but one of the 41 first antibody-positive specimens (97.6%) were PCR positive and 65% were isolation positive. Two control specimens from seronegative homosexual men were PCR positive and one was culture positive, but HLA typing provided clear evidence of specimen mix-up in the specimen archive. For 37/41 seroconvertors, all available antibody negative specimens were negative by both PCR and virus isolation. In three cases, the specimen obtained 6 months before seroconversion was PCR and isolation positive. One specimen, obtained 12 months before SC, was PCR positive and isolation negative but was determined to be the result of sample contamination. Both PCR and isolation were negative in this subject 6 months before SC. In conclusion, we were unable to detect immunosilent infection >6 months before seroconversion in high-risk homosexual men.