Because of the extremely short photochemical lifetime of tropospheric OH, comparisons between observations and model calculations should be an effective test of our understanding of the photochemical processes controlling the concentration of OH, the primary oxidant in the atmosphere. However, unambiguous estimates of calculated OH require sufficiently accurate and complete measurements of the key species and physical variables that determine OH concentrations. The Tropospheric OH Photochemistry Experiment (TOHPE) provides an extremely complete set of measurements, sometimes from multiple independent experimental platforms, that allows such a test to be conducted. When the calculations explicitly use observed NO, NO2, hydrocarbons, and formaldehyde, the photochemical model consistently overpredicts in situ observed OH by ~50% for the relatively clean conditions predominantly encountered at Idaho Hill. The model bias is much higher when only CH4-CO chemistry is assumed, or NO is calculated from the steady state assumption. For the most polluted conditions encountered during the campaign, the model results and observations show better agreement. Although the comparison between calculated and observed OH can be considered reasonably good given the ¿30% uncertainties of the OH instruments and various uncertainties in the model, the consistent bias suggests a fundamental difference between theoretical expectations and the measurements. Several explanations for this discrepancy are possible, including errors in the measurements, unidentified hydrocarbons, losses of HOx to aerosols and the Earth's surface, and unexpected peroxy radical chemistry. Assuming a single unidentified type of hydrocarbon is responsible, the amount of additional hydrocarbon needed to reduce theoretical OH to observed levels is a factor of 2 to 3 greater than the OH-reactivity-weighted hydrocarbon content measured at the site. Constraints can be placed on the production and yield of various radicals formed in the oxidation sequence by considering the observed levels of certain key oxidation products such as formaldehyde and acetaldehyde. The model results imply that, under midday clean westerly flow conditions, formaldehyde levels are fairly consistent with the OH and hydrocarbon observations, but observed acetaldehyde levels are a factor of 4 larger than what is expected and also imply a biogenic source. Levels of methacrolein and methylvinylketone are much lower than expected from steady state isoprene chemistry, which implies important removal mechanisms or missing information regarding the kinetics of isoprene oxidation within the model. In a prognostic model application, additional hydrocarbons are added to the model in order to force calculated OH to observed levels. Although the products and oxidation steps related to pinenes and other biogenic hydrocarbons are somewhat uncertain, the addition of a species with an oxidation mechanism similar to that expected from C10 pinenes would be consistent with the complete set of observations, as opposed to naturally emitted isoprene or any of the anthropogenic hydrocarbons examined in the model. Further constraints on the abundance of peroxy radicals are necessary in order to fill the gaps in our understanding of OH photochemistry for the clean continental conditions typical of Idaho Hill.¿ 1997 American Geophysical Union