We examined the effect of state and federal legislative actions on the incidence of human salmonellosis associated with pet turtles. Using the Salmonella Surveillance System of the Center for Disease Control, we identified Salmonella serotypes that were and were not specifically turtle associated. From 1970 to 1976, there was a 77% reduction in frequency of turtle-associated serotypes and no consistent trend among non-turtle-associated serotypes. Decrease in frequency of turtle-associated serotypes correlated best with decrease in total production of pet turtles. Reduction in turtle-associated sertotypes temporally correlated with an 18% reduction in frequency of salmonellosis in children, aged 1 to 9 years. Frequency of turtle-associated serotypes rapidly fell in states importing turtles, in contrast to states producing them. Continued although infrequent occurrence of salmonellosis in children with pet turtles counsels against relaxing legislation concerned with its control.