OBJECTIVE: To evaluate variations in work-loss estimates by phone interview method. METHODS: In phone interviews, total work loss estimates were based on three domains: missed workdays; missed hours, and reduced productivity on days at work while not feeling well. Three different phone interviews were developed. Version 1 (V1) included a lengthy direct assessment of work loss. Version 2 (V2) was an abridged version of V1. Version 3 (V3) included a brief indirect assessment of work loss. Two recall periods at one week and at four weeks were also tested. Combining the three versions and the two recall periods yielded six different interviews. A convenience sample (n = 20,088) of adult residents from the Baltimore, MD and Chicago, IL areas was contacted by phone, of whom 7,691 met occupation eligibility criteria. Respondents were randomized to one of the six interviews at the time of contact. Interviews were compared on the basis of estimates of work loss and variation in estimates of work loss associated with recall period and version. RESULTS: Interviews varied in average length from 9.2 minutes (V3) to 15.1 minutes (V1). The participation rate (66%) did not vary with the version. Estimates of work loss varied with recall period and version. The mean total hours/week of lost productive work time was lower for interviews using a four-week compared to a one-week recall period. V3 of the interviews was the most discriminating for illness-related work-loss estimates. For example, individuals with allergic rhinitis (AR) reported significantly more lost productive time at work due to not feeling well compared to controls and a stronger relation between severity of AR and work loss. CONCLUSION: Variation in relatively short recall periods influences estimates of work loss. Interview version 3 required the least time to administer and was the most discriminating in terms of work-loss estimates by illness status and illness severity.