Interpolation methods can estimate individual-level exposures to air pollution from ambient monitors; however, few studies have evaluated how different approaches may affect health risk estimates. We applied multiple methods of estimating exposure for several air pollutants. We investigated how different methods of estimating exposure may influence health effect estimates in a case study of lung function data, forced expiratory volume in 1 s (FEV1), and forced vital capacity (FVC), for 2102 cohort subjects in Ulsan, Korea, for 2003–2007. Measurements from 13 monitors for particulate matter <10 μm (PM10), ozone, nitrogen dioxide, sulfur dioxide, and carbon monoxide were used to estimate individual-level exposures by averaging across values from all monitors, selecting the value from the nearest monitor, inverse distance weighting, and kriging. We assessed associations between pollutants and lung function in linear regression models, controlling for age, sex, and body mass index. Cross-validation indicated that kriging provided the most accurate estimated exposures. FVC was associated with all air pollutants under all methods of estimating exposure. Only ozone was associated with FEV1. An 11 ppb increase in lag-0–2 8-h maximum ozone was associated with a 6.1% (95% confidence interval 5.0, 7.3%) decrease in FVC and a 0.50% (95% confidence interval 0.03, 0.96%) decrease in FEV1, based on kriged exposures. Central health effect estimates were generally higher using exposures based on averaging across all monitors or kriging. Results based on the nearest monitor approach had the lowest variance. Findings suggest that spatial interpolation methods may provide better estimates than monitoring values alone by reflecting the spatial variability of individual-level exposures and generating estimates for locations without monitors.