Epithelial-cell micronuclei (MN) are potentially useful markers of occupational exposure to genotoxicants. With intermittent exposures, cells sampled either before or after a specific time interval, reflecting the time it takes for damaged cells to become available at the epithelial surface, are unlikely to be exposure-related. It may then be important to conduct an exposure-window analysis, with the goal of identifying the relevant exposures.We re-analysed individual exposure data from a previous study (Suruda et al. 1993) of MN formation in 22 male mortuary science students exposed to formaldehyde during a 90-day embalming class. We conducted an exposurewindow analysis and compared the results with those obtained with 90-day cumulative exposure. The window widths varied between 7 and 25 days, in 1 day increments, assuming a constant 7-day cell-cycle. We assessed the fit (likelihood-ratio test) of a linear regression model, regressing the change in buccal MN prevalence on formaldehyde exposure, using both asymptotic and non-asymptotic methods. Exposures defined from 7-15 to 7-18 days before specimen collection provided a slightly better fit than the 90-day cumulative exposure, with a doubling of the regression coefficient for the exposure effect (for the 7-16-days window LR = 5.32, p = 0.032, coefficient = 0.088 MN per 1000 cells per ppm-hr; 95% CI = 0.014, 0.16; for the 90-day cumulative exposure LR = 4.44, p = 0.048, coefficient = 0.045 MN per 1000 cells per ppm-hr, 95% CI = 0.0038, 0.086). Although hampered by the small number of subjects, these results reinforce the potential importance of exposure timing.