This case-control study examines potential associations between telephone linework and the occurrence of leukemia except chronic lymphocytic leukemia in a primarily retired population of American Telephone and Telegraph Company (AT&T) workers. Cases died between 1975 and 1980. Exposure is defined both by job title and, for workers with complete job histories, by a lifetime exposure score based on industrial hygiene personal monitoring measurements of line and nonline jobs. When the time-weighted average mean for each job is accumulated into a lifetime exposure score, workers with scores above the median for the population show an excess of leukemia 2.5 times higher than workers below the median (95% confidence interval (CI) 0.7-8.6). Those individuals with long duration of employment in jobs with intermittent peak exposures may be at higher risk of leukemia than those with a constant exposure level. Analyses that allow for a latent period suggest the risk is associated with exposures that occurred 10 or more years before death. Workers with peak exposure scores above the median have odds ratios of 2.4 (95% CI 0.7-9.0) and 6.6 (95% CI 0.7-58) for latent periods of 10 and 15 years, respectively. The data suggest an increasing risk with increasing exposure (p for trend = 0.05) when cumulated scores are based on peak exposure scores. Peak exposures tended to occur in cable splicing work and in old telephone switching offices. The numbers in this study are small and observed differences may be due to chance.