Abstract A randomized, placebo-controlled trial of the efficacy of pyrantel (single dose 12·5 mg/kg bodyweight), mebendazole (single 500 mg dose) and albendazole (single 400 mg dose) in the treatment of hookworm infections ( Necator americanus) was carried out in January 1998 in the Southern Region of Mali, West Africa, during the period of Ramadan (Islamic fast). Statistical analysis of the pre-intervention faecal egg counts showed that there was a significant pre-treatment chance bias, despite randomization of subjects into treatment groups, arising from the main effect of sex (heavier infections among males) and a sex × treatment interaction (the sex bias was not evident in the pyrantel-treatment group). The participants were reexamined 10 days after treatment, and after controlling for the drift in faecal egg counts in the placebotreated subset, age, sex, fasting and intensity of infection, albendazole was clearly the most effective drug showing consistently efficacies in the range 92·1 to 99·7%, depending on the method of evaluation and the particular subset of the treatment group. Neither mebendazole nor pyrantel was as effective, with efficacies ranging from 60·9 to 89·8% and 4·8 to 89·7%, respectively. Fasting made no difference to drug efficacy. On the basis of our results the single 400 mg dose of albendazole is the treatment of choice for hookworm infections in this region of Mali. We emphasize the need for standardization of the methods used for trial designs, for calculation of summary data relating to drug efficacies and the accompanying statistical tests.