Data on measles from the project in the Machakos district, Kenya between 1974 and 1981 have been reanalysed in order to test the impact of sex and cross-sex transmission on severity of infection. In families with several cases, the case fatality rate was high as 11.3% (13/115) during the initial 6 months of the project. In the remaining period, the case fatality rate fell to 2.4% (21/885) (relative risk (RR) = 0.21, 95% confidence interval (CI): 0.11-0.39). During the initial period with high mortality, second cases had 4.74 times higher mortality (95% CI: 1.65-13.66) than index cases and there was no difference in mortality between girls and boys (RR = 0.98). Among secondary cases, though not significant, those infected by someone of the opposite sex had a trend toward a higher risk of dying than those infected by someone of their own sex (RR = 2.44, 95% CI: 0.77-7.78). In families with two children of the same sex, the case fatality rate was 9% compared with 29% in families with a boy and a girl (RR = 3.49; 95% CI: 0.96-12.75). In the subsequent period with low mortality, the difference in mortality between index and secondary cases was less pronounced (RR = 2.32, 95% CI: 1.03-5.25) and girls had significantly higher case fatality than boys (RR = 2.63, 95% CI: 1.09-6.34). There was no difference in this case fatality rate associated with cross-sex transmission of infection (RR = 0.88).(ABSTRACT TRUNCATED AT 250 WORDS)