In order to assess the possible human carcinogenicity of formaldehyde we conducted a retrospective cohort mortality study of workers exposed for at least three months to formaldehyde in three garment facilities which produced permanent press garments. A total of 11,030 workers contributing 188,025 person-years were included in the study. Vital status was successfully ascertained through 1982 for over 96% of the cohort. The average (TWA) formaldehyde exposure at the three plants monitored in 1981 and 1984 by NIOSH was 0.15 ppm but past exposures may have been substantially higher. In general, mortality from nonmalignant causes was less than expected. A statistically significant excess in mortality from cancers of the buccal cavity (SMR = 343) and connective tissue (SMR = 364) was observed. Statistically nonsignificant excesses in mortality were observed for cancers of the trachea, bronchus and lung (SMR = 114), pharynx (SMR = 112), bladder (SMR = 145), leukemia and aleukemia (SMR = 113), and other lymphopoietic neoplasms (SMR = 170). Mortality from cancers of the trachea, bronchus and lung was inversely related to duration of exposure and latency. In contrast, mortality from cancers of the buccal cavity, leukemias, and other lymphopoietic neoplasms increased with duration of formaldehyde exposure and/or latency. These neoplasms also were found to be highest among workers first exposed during a time period of high potential formaldehyde exposures in this industry (1955-1962). However, it should be recognized that these findings are based on relatively small numbers and that confounding by other factors may still exist. The results from this investigation, although far from conclusive, do provide evidence of a possible relationship between formaldehyde exposure and the development of upper respiratory cancers (buccal), leukemias, and other lymphopoietic neoplasms in humans.