It has been suggested, based on animal experiments and limited human data, that the anti-hypertensive drug hydralazine might be carcinogenic, and among the sites of concern are the lung and colon. To assess the possible relationship between the use of hydralazine and lung and colorectal cancers in humans, we compared 1006 cases of lung cancer with 3531 hospital control subjects, and 972 cases of colorectal cancer with 3276 controls. Data were collected by interview in hospitals in the United States and Canada. Overall, 1.1% of the lung cancer cases, 1.6% of the colorectal cancer cases, and 1.5% of the controls had used hydralazine. Compared with those who had never used hydralazine, the relative risk estimate of lung cancer for those who first took the drug at least 18 months before hospital admission was 1.1 (95% confidence interval 0.4-2.9). The estimate for use for at least 1 year was 1.4 (0.5-3.8) and for use for at least 5 years the estimate was 0.9 (0.2-4.3). The corresponding relative risk estimates for colorectal cancer were 1.2 (0.5-2.5) for any use, 1.7 (0.8-3.7) for use for at least one year, and 2.4 (0.8-6.9) for five or more years of use. Other antihypertensive treatments and risk factors, including cigarette smoking in the analysis of lung cancer, were taken into account in these estimates. Although the effect of use after long latent intervals could not be evaluated, the results provide no support for the hypothesis that hydralazine use increases the risk of lung cancer.(ABSTRACT TRUNCATED AT 250 WORDS)