The effect of control of houseflies on the incidence of diarrhoea and shigellosis was evaluated in a prospective crossover intervention study at two military field bases several kilometers apart. In early summer, 1988, intensive fly control measures (mainly bait and trap strategy) were introduced on one base, while the other served as a control. After 11 weeks, as new cohorts arrived, the intervention was abruptly discontinued in the first base and instituted in the second for the next 11 weeks. The study was repeated the next summer. Overall, fly counts were 64% lower on the bases exposed to fly control measures (p = 0.024). Concomitantly, clinic visits dropped by 42% (p = 0.146) for diarrhoeal diseases and by 85% for shigellosis (p = 0.015), as did rates of seroconversion, by 76% (p = 0.024) for antibodies to Shigella and by 57% (p = 0.006) for antibodies to enterotoxigenic Escherichia coli. The findings indicate that houseflies, acting as mechanical vectors, transmit Shigella (and possibly enterotoxigenic E coli) diarrhoeal infections.