Cancer morbidity and mortality were examined among 11,894 Puerto Rican participants in a US Public Health Service preventive therapy trial begun in 1957. The mean duration of follow-up was 18 years (range 16-19 years). No statistically significant difference in overall cancer rates was observed between the participants in the trial assigned isoniazid and those assigned placebo. There were also no significant differences between the groups when rates for specific types of cancer and age-specific cancer rates were compared. No trend toward an increasing cancer rate in the isoniazid group was observed with the passage of time. The mortality rate due to cancer was similar in the two groups. These data coupled with other information reported in the literature provide substantial evidence that isoniazid, when given in the usual therapeutic doses, is not carcinogenic in humans.