Understanding behavioral responses to changes in actual or perceived risk is important because risk-reduction goals can be undermined by risk-compensating behavior. This paper examines the response to new information about the risk of HIV infection. Approximately 1,200 circumcised and uncircumcised men in rural Malawi are randomly informed that male circumcision reduces the HIV transmission rate, predicting asymmetric behavioral responses. We find no evidence that the information induces circumcised men to engage in riskier sex while uncircumcised men practice safer sex in response to the information. There were no significant effects of the information on child circumcisions after one year.