A detailed modeling analysis is conducted focusing on nonmethane hydrocarbons and ozone in three southeast United States national parks for a 15-day time period (14--29 July 1995) characterized by high O3 surface concentrations. The three national parks are Smoky Mountains National Park (GRSM), Mammoth Cave National Park (MACA), and Shenandoah National Park (SHEN), Big Meadows. A base emission scenario and eight variant predictions are analyzed, and predictions are compared with data observed at the three locations for the same time period. Model-predicted concentrations are higher than observed values for O3 (with a cutoff of 40 ppbv) by 3.0% at GRSM, 19.1% at MACA, and 9.0% at SHEN (mean normalized bias error). They are very similar to observations for overall mean ozone concentrations at GRSM and SHEN. They generally agree (the same order of magnitude) with observed values for lumped paraffin compounds but are an order of magnitude lower for other species (isoprene, ethene, surrogate olefin, surrogate toluene, and surrogate xylene). Model sensitivity analyses here indicate that each location differs in terms of volatile organic compound (VOC) capacity to produce O3, but a maximum VOC capacity point (MVCP) exists at all locations that changes the influence of VOCs on O3 from net production to production suppression. Analysis of individual model processes shows that more than 50% of daytime O3 concentrations at the high-elevation rural locations (GRSM and SHEN) are transported from other areas; local chemistry is the second largest O3 contributor. At the low-elevation location (MACA), about 80% of daytime O3 is produced by local chemistry and 20% is transported from other areas. Local emissions (67--95%) are predominantly responsible for VOCs at all locations, the rest coming from transport. Chemistry processes are responsible for about 50% removal of VOCs for all locations; less than 10% are lost to surface deposition and the rest are exported to other areas. Metrics, such as VOC potential for O3 production (VPOP), which links the chemistry processes of both O3 and VOCs and MVCP, are devised to measure the different characteristics of O3 production and VOCs. The values of the defined metrics are mapped for the entire modeling domain. Implications of this model exercise in understanding O3 production are analyzed and discussed. Even though this study was focused on three United States national parks, the research results and conclusions may be applicable to other or to similar rural environments in the southeast United States.