Occupational physicians investigate perceived cancer clusters to alleviate employee concerns and pursue etiologic hypotheses. We conducted a retrospective cohort analysis of all past and present employees of a metal fabrication plant and a comparison plant after employees recognized five cancer cases in 1987. We ascertained cases of all subjects who were employed at some time in the 8 years before 1987 through the Colorado Central Cancer Registry and determined vital status through the National Death Index. Cancer incidence at the index plant was almost identical to that of the population of the Denver metropolitan area (standardized incidence ratio [SIR] = 99, 95% confidence interval [CI] 59-165). Proportional incidence ratios revealed that no type of cancer occurred with significant excess in the index plant population during 1979 through 1986. Where population-based tumor registries exist, occupational physicians can employ this inexpensive and robust methodology to assess cancer incidence in exposed cohorts, pursue exposure-response relations, and evaluate clusters.