The effects of recall on self reporting of occupational injuries in epidemiological studies were examined. Data on occupational injuries were taken from the 1988 Occupational Health Supplement (OHS) to the National Health Interview Survey (NHIS). The OHS collected data on occupational injuries and illnesses from one randomly selected adult in each household that participated in the NHIS. This yielded 29,412 persons who had worked at some time during the past year who were interviewed to obtain information on any injuries they had suffered at work during the year. The data were used to obtain unadjusted annual injury rates. A linear model was fitted to the unadjusted injury rates to estimate the incidence rate expected if all respondents had been interviewed within 4 weeks of their injury (adjusted recall injury rate). The overall unadjusted rate of injury was 8.55 injuries per 100 workers per year (yr). The adjusted recall injury rate was 11.29/100/yr, which represented a 32.0% increase over the unadjusted rate. By sex, the unadjusted rates were: males 11.05/100/yr and females 5.61/100/yr. The adjusted recall injury rates were: males 14.4/10/yr and females 7.62/100/yr. By job class, the unadjusted rates for blue and white collar workers were 17.23 and 4.5/100/yr, respectively. The adjusted recall rates for blue and white collar workers were 22.89 and 5.86/100/yr, respectively. The unadjusted rates for injuries resulting in 1 or more lost workdays and no lost workdays were 3.92 and 4.26/100/yr, respectively. The corresponding adjusted recall rates were 4.81 and 6.10/10/yr. The authors conclude that using a 12 month or longer reference period in injury surveys results in an underestimate of work related injuries. A shorter reference period such as 4 weeks improves recall and is desirable for more accurate estimates. To use a shorter reference period requires increasing the size of the population to be sampled, therefore increasing the cost of the survey. The increased cost must be balanced against the need to obtain more accurate injury information.