A case-referent study was undertaken to investigate the associations between formaldehyde exposure and respiratory and related cancers. Fifty-seven such cancers from a retrospective cohort of male woodworkers formed the case group. They were matched by year of birth with 171 referents. Exposure to formaldehyde was assessed with job-exposure matrices. The median of the time-weighted average concentration was about 1 ppm, and the mean duration of exposure was 10 years among the exposed. Odds ratios (OR) were calculated for formaldehyde exposure (1.44), peak exposure to formaldehyde (1.26), and exposure to formaldehyde-containing wood dust (1.22). None of the values exceeded unity with statistical significance. Allowance was also made for a 10-year period from the onset of exposure. Birth year, cigarette smoking and exposure to wood dust, chlorophenols, pesticides, and terpenes were controlled by stratification. The adjusted ORs did not change appreciably. The highest OR was 1.95 for formaldehyde exposure without allowance for minimum latency, adjusted for exposure to terpenes. No exposure-response relation was observed for the level, duration, or dose (ppm-years) of formaldehyde exposure. The result is nonpositive and may be explained by absence of effect, by too short a follow-up, or by insufficient power for detecting a mild excess risk.