The reproducibility of topical application data is influenced by test duration and the definition of death. These parameters were varied in tests with fall armyworms,Spodoptera frugiperda(J. E. Smith), to determine conditions which gave the most uniform results. Assessments where moribund insects are considered dead most closely approximate ‘actual’ mortality (i.e. at 7 days both methods gave equivalent results). With this criterion, reliable data can be obtained within 48 h. We recommend combining the ESA guidelines for test duration with the FAO guidelines for evaluating mortality. A method for analyzing the reliability of probit data from repeated tests, where the same insects are evaluated under different mortality criteria, is developed.