Ozone (O ) damage to leaves can reduce plant photosynthesis, which suggests that declines in ambient O concentrations ([O ]) in the United States may have helped increase gross primary production (GPP) in recent decades. Here, we assess the effect of long-term changes in ambient [O ] using 20 years of observations at Harvard forest. Using artificial neural networks, we found that the effect of the inclusion of [O ] as a predictor was slight, and independent of O concentrations, which suggests limited high-frequency O inhibition of GPP at this site. Simulations with a terrestrial biosphere model, however, suggest an average long-term O inhibition of 10.4% for 1992-2011. A decline of [O ] over the measurement period resulted in moderate predicted GPP trends of 0.02-0.04 μmol C m s yr , which is negligible relative to the total observed GPP trend of 0.41 μmol C m s yr . A similar conclusion is achieved with the widely used AOT40 metric. Combined, our results suggest that ozone reductions at Harvard forest are unlikely to have had a large impact on the photosynthesis trend over the past 20 years. Such limited effects are mainly related to the slow responses of photosynthesis to changes in [O ]. Furthermore, we estimate that 40% of photosynthesis happens in the shade, where stomatal conductance and thus [O ] deposition is lower than for sunlit leaves. This portion of GPP remains unaffected by [O ], thus helping to buffer the changes of total photosynthesis due to varied [O ]. Our analyses suggest that current ozone reductions, although significant, cannot substantially alleviate the damages to forest ecosystems.