We assessed the ability of radioimmunoassay for hepatitis B e antigen (HBeAg) to predict infectivity in exposed medical personnel by analyzing 390 samples of sera positive for hepatitis B surface antigen (HBsAg) that were implicated in accidental inoculations of known outcome. The radioimmunoassay detected HBeAg or its antibody (anti-HBe) in 91% of the donor sera. The incidence of hepatitis B was 19% (44 of 234) in recipients of HBeAg-positive sera but was only 2.5% (three of 121) in recipients of sera positive for anti-HBe, and nil (none of 35) in recipients of sera negative for HBeAg and anti-HBe. The known relation of HBeAg and infectivity was quantified by radioimmunoassay as a risk ratio of 10:1 (HBeAg-positive to HBeAg-negative) for this type of exposure. The sensitivity of the radioimmunoassay also showed that a large proportion (55%) of donor sera not producing hepatitis were positive for HBeAg; therefore, even the most flagrant needlestick exposures to HBsAg-positive sera often must involve subthreshold amounts of infective material.