Longitudinal change in FEV1(deltaFEV1) has been useful in assessing adverse effects of exposures and/or disease processes. Differences between estimated FEV1, slope (capped beta) and "true" slope (beta), resulting from biological variation and measurement errors, expressed by the "error' term (e=l capped beta - betal), can be minimized by increasing number of subjects (N), years of follow-up (D), and/or frequency of measurements (P). To evaluate different strategies for study design, we compared the maximum error values (emax) using actual data obtained from 160 coal miners and working non-miners who had performed 11 spirometry measurements using ATS standards at 6-mo intervals over 5 years. Individual 5-yr deltaFEV1 were calculated by linear regression We determined emax by setting the probability that error is less than or equal to emax to be 95% That is P[ l capped beta - beta l < / = emax]=0.95. For the longer study durations, testing either every 6 or 12months resulted In similar errors (4 yrs - 10 vs. 16 ml/yr,5 yrs - 8vs. 11ml/yr). For 2 and 3 yr studies, errors were 4.4 and 2 times higher for 12 vs 6 monthly measurements. When designing longitudinal spirometry studies, the use of actual field test results can assist in defi8ning available options to achieve a desired value of maximum error.