The microdroplet lymphocyte cytotoxicity test was examined thoroughly in an effort to increase the reproducibility of the test. The discrepancy rate in a large series of tests was reduced from 5.16% at the start of this study to the present 0.95% by introducing certain modifications in the technique. Variables connected with the isolation of lymphocytes, handling of antisera, quality of antisera, amount of complement, incubation temperature, duration of incubation, fixing of reactions, and reading of reactions were studied. The method which has resulted appears to be reproducible, simple, and readily usable on a large scale.