Techniques for examining the consistency of results reported in epidemiologic and toxicologic studies were described and illustrated using methylene-chloride (75092) (MC). Data were taken from the mouse component of a longterm inhalation bioassay of MC toxicity and carcinogenicity in B6C3F1-mice and F344/N-rats by the National Toxicology Program (NTP) and a study of lung and liver cancer mortality among Kodak company workers exposed to MC vapor (Kodak study). The NTP study found that MC induced liver and lung cancer in mice and mammary tumors in rats. No increase in mortality from lung or liver cancer was found in the Kodak study. The techniques involved adjusting for different lengths of followup in humans versus mice using a multistage carcinogenesis risk assessment model, defining an appropriate dose metric that would produce equivalent risks for humans and mice, and calculating standardized mortality ratios (SMRs) by applying the multistage carcinogenesis model to the mouse bioassay data after adjusting the Kodak study data for the healthy worker effect. Confidence intervals (CIs) were computed for the SMRs obtained in the Kodak study and compared with predicted CIs derived from applying the multistage carcinogenesis model to the mouse bioassay data. Adjusting for the healthy worker effect decreased the number of expected deaths from lung and liver cancer in the Kodak study; however, none of the SMRs exceeded unity. The CIs of the SMRs predicted by applying the multistage carcinogenesis model to the mouse data were clearly nested within those obtained in the Kodak study. The authors conclude that the negative results obtained in the Kodak study are not inconsistent with predictions from the mouse bioassay data after adjusting for differences in study protocol, the healthy worker effects, and statistical variability.