Ambient monitors are commonly used to estimate exposure for epidemiological studies, and air quality modeling is infrequently applied. However air quality modeling systems have the potential to alleviate some, although not all, of the limitations of monitoring networks. To investigate this application, exposure estimates were generated for a case study high ozone episode in the Northern Georgia Region of the United States based on measurements and concentration estimates from an air quality modeling system. Hourly estimates for 2268 4-km by 4-km gridcells were generated in a domain that includes only eight ozone monitors. Individual and population-based ozone exposures were estimated using multiple approaches, including area-weighted average of modeled estimates, nearest monitor, and spatial interpolation by inverse distance weighting and kriging. Results based on concentration fields from the air quality modeling system revealed spatial heterogeneity that was obscured by approaches based on the monitoring network. With some techniques, such as spatial interpolation, monitoring data alone was insufficient to estimate exposure for certain areas, especially for rural populations. For locations far from ozone monitors, the estimates from the nearest monitor approach tended to overestimate exposure, compared to modeled estimates. Counties in which one or more monitors were present had statistically higher population density and modeled ozone estimates than did counties without monitors (p-value < 0.05). This work demonstrates the use of air quality modeling to generate higher spatial and temporal resolution exposure estimates, and compares the advantages of this approach to traditional methods that use monitoring data alone. The air quality modeling method faces its own limitations, such as the need to thoroughly evaluate concentration estimates and the use of ambient levels rather than personal exposure.