SUMMARY Using the regression model proposed by Cox (1972) a test for monotone trend in hazard functions is derived for use, for example, in chemical carcinogenesis experiments in which a chemical is administered at increasing dose levels. It is demonstrated that under the null hypothesis of no dose-response relationship the test statistic is distributed asymptotically as chi-squared with one degree of freedom. An exact test for trend is proposed which is valid when censorship is random and equal.