This study examined the morbidity experience of a prospective cohort of 2132 male employees who worked at a petroleum refinery from 1981 through 1988. The morbidity data included all illness-absence episodes in excess of 5 days during the study period. Standardized morbidity ratios (SMRs) of disease prevalence were calculated using data from all manufacturing employees of the Shell Oil Company as an internal comparison group. As such, there is no potential bias associated with the "healthy worker effect" in this type of study design. Morbidity for all causes combined was virtually the same as that for the comparison group with 2,311 observed and 2,318 expected disease prevalence events. However, there were statistically increased prevalence of musculoskeletal system disorders (SMR = 136) and injuries (SMR = 125) among staff employees and skin and subcutaneous tissue disorders (SMR = 138) among production employees. A review of the original morbidity reports for these skin conditions revealed that none were due to exposure to chemical products or solvents. The SMR for neoplasms of the lymphatic and hematopoietic tissue among production employees was slightly elevated but was based on only three cases (2.4 expected). Of the three cases, none was due to leukemia.