This study of the effect on CD4%, CD8%, CD3+8+%, and CD3% of flow cytometer, monoclonal antibody, and fluorochrome was based on 71 whole-blood samples, each evaluated by 42 to 59 laboratories during 2 years of a flow cytometry quality assessment program. For the 24 HIV-positive specimens, FACScans produced significantly lower CD4% values than EPICS-Cs or EPICS Profiles, and for the 47 HIV-negative specimens, FITC was associated with significantly lower CD4% values than PE or RD1, but differences were never larger than 2% and regressions accounted for only 3-12% of the variability. The labs using the most common CD4 technique had significantly higher between-laboratory variability than all other labs grouped together. For both CD8 and CD3+8+, measurements on FACScans were significantly higher than measurements on EPICS, and measurements using Leu2 were significantly higher than measurements using T8, with regressions accounting for 12-31% of the variability. The machine differences in medians were 3-7% for labs using Leu2-FITC. It might be worthwhile to discourage the use of Leu2-FITC for measuring CD8% but no change in instrument, monoclonal antibody, or fluorochrome would greatly improve interlaboratory agreement on CD4%.