OBJECTIVES: To evaluate the feasibility and validity of daily self-reported performance in relation to objective productivity data, in order to develop measures of health effects on work function. METHODS: Medical bill reviewers at 3 worksites of the same firm provided daily information through interactive voice response (IVR) on speed, concentration and accuracy compared to their best on a 1-10 ordinal scale for 12 weeks. The 7-item Work Limitations Questionnaire (WLQ) was administered monthly. Objective data included types of bills processed and specific processing activities. Weights for relative difficulty of activity were applied, and outliers were removed. RESULTS: One hundred and twenty four of 142 recruited subjects (87%) completed > 50% of daily IVR reports. Concentration, speed and accuracy ratings were highly inter-correlated (R=0.75), and right-skewed (mean speed=7.7, SD=1.5). The mean adjusted productivity rate (MAP) was 34 bills/hour (range 4.7 to 111, SD12.6, 61% within-person variation). Subject-specific speed/MAP correlation ranged from R=-.20 to .75 (mean, .28). Health status, years on job, age, IVR completion rate, month of study, or total hours worked were not associated with variations across individuals in R. Correlations among WLQ responses, IVR speed, and monthly objective productivity were low (R<0.12). CONCLUSIONS: Daily subjective and objective data collection are feasible; both may be equally valid, in the sense that they may be measuring different aspects of work performance. Even with a relatively well-defined job and limited range of tasks, the "objective" data was complex and challenging to analyze. Low subjective/objective correlations may be due to respondent inability to relate productivity to the IVR questions or monthly survey items; subjects incorporating difficulty adjustment in their ratings and/or failure of 'objective' data to capture the full complexity of the job or difficulty variation.