Reported efficacies of drugs used to treat Strongyloides stercoralis infection vary widely. Because diagnostic methods are insensitive, therapeutic trials generally require multiple negative posttreatment stool specimens as evidence of drug efficacy. However, only a single positive stool specimen is usually required for study enrollment. To determine the reproducibility of detection of S. stercoralis larvae in the stool, 108 asymptomatic infected men submitted 25 g of fresh stool once a week for eight consecutive weeks for examination by the Baermann technique. During the 8-week study, 239 (27.7%) of 864 stool specimens were positive for S. stercoralis. Rates of detection of larvae in the stool specimens ranged from eight of eight specimens in 3 (2.8%) men to none of eight specimens in 36 (33.3%) men. Of 43 men for whom S. stercoralis was detected in at least two of the first four stool specimens, only 1 (2.3%) man tested negative on all of the next four specimens. In comparison, of 29 men who had detectable larvae in only one of the first four specimens, 22 (75.9%) tested negative on all of the next four samples. Thus, if these 29 men had been enrolled in a therapeutic trial between the first and second sets of four specimens, the efficacy of a drug with no activity against this parasite would have been estimated to be 76%. These data suggest that patterns of S. stercoralis detection vary widely among infected persons and that intermittent larval shedding can lead to inflated estimates of drug efficacy. Before a patient is entered in a clinical trial of drug efficacy, four consecutive stool specimens should be examined for S. stercoralis; only persons with two or more positive specimens should be enrolled.