Hydroxyl (OH) and hydroperoxyl (HO2) strongly influence the atmosphere's oxidation of gases emitted from Earth's surface and the formation and aging of aerosol particles. Thus, understanding OH and HO2 chemistry is essential for examining the impact of human activity on future atmospheric composition and climate. Using the OH and HO2 data set collected with the Penn State Airborne Tropospheric Hydrogen Oxides Sensor (ATHOS) during nine aircraft missions over the past 20 years, we compare observed OH and HO2 to that modeled using the same near-explicit photochemical box model. In general, the agreement is well within the combined uncertainties of the observations and models, even when the model is constrained only with a common data set of simultaneous measurements. However, in environments such as cities, forests, and pollution plumes, the model chemical mechanism and size of the data set of constraining measurements do matter. The model constrained by the full set of measurements better simulates OH loss and OH in these regions than the model constrained by the common set of measurements. In cleaner regions, the differences between observed and modeled OH and HO2 found in previous studies generally remain and do not appear to be systematic, indicating that the differences are driven by measurement issues for ATHOS and/or other instruments. Thus, these comparisons indicate that the oxidation chemistry in most of the free troposphere is understood to as well as current measurements can determine. The focus of future research needs to be on regions rich in volatile organic compounds, where observed-to-modeled differences are more persistent, and on improving measurement consistency. Plain Language Summary Understanding the atmospheric chemistry that results in the removal of important greenhouse gases, such as methane, is critical in assessing the impact of human activity on future climate. Using airborne atmospheric data collected over a twenty-year period, we compare observed and modeled hydroxyl and hydroperoxyl, two important reactive gases, to evaluate our ability to model this oxidation chemistry. Overall, these comparisons show agreement between observed and modeled hydroxyl and hydroperoxyl to within combined instrument and model uncertainties although the level of agreement degrades over polluted or forested regions and is variable from mission to mission and even within missions. The lack of systematic disagreement between observed and modeled hydroxyl and hydroperoxyl when using the same model for all the studies suggests that unexpected changes in the performance of multiple instruments are likely responsible for this variable level of agreement.