Results from a tightly constrained photochemical point model for OH and HO2 are compared to OH and HO2 data collected during the Program for Research on Oxidants: Photochemistry, Emissions, and Transport (PROPHET) summer 1998 intensive campaign held in northern Michgian. The PROPHET campaign was located in a deciduous forest marked by relatively low NOx levels and high isoprene emissions. Detailed HOx budgets are presented. The model is generally unable to match the measured OH, with the observations 2.7 times greater than the model on average. The model HO2, however, is in good agreement with the measured HO2. Even with an additional postulated OH source from the ozonolysis of unmeasured terpenes, the measured OH is 1.5 times greater than the model; the model HO2 with this added source is 15% to 30% higher than the measured HO2. Moreover, the HO2/OH ratios as modeled are 2.5 to 4 times higher than the measured ratios, indicating that the cycling between OH and HO2 is poorly described by this model. We discuss possible reasons for the discrepancies.