A global three-dimensional model of tropospheric O3-NOx-hydrocarbon chemistry is used to investigate the factors controlling ozone concentrations in the troposphere. Model results indicate a close balance between chemical production and chemical loss of ozone in the tropospheric column at all latitudes (except high latitudes in winter). Using separate tracers for ozone produced in the stratosphere and in different regions of the troposphere, we find that the contribution of transport from the stratosphere to ozone concentrations in the troposphere is about 30% at midlatitudes in winter, 10% in summer, and 5% in the tropics. Production of ozone in the upper, middle, and continental lower troposphere all make significant contributions (10–50%) to ozone concentrations throughout the troposphere. The middle troposphere is a major global source region for ozone even though it is not a region of net production. The springtime maximum of ozone observed at remote sites in the northern extratropics is explained by a phase overlap between ozone transported from the stratosphere which peaks in late winter and ozone produced in the troposphere which peaks in late spring. Our model results do not support previous explanations of the springtime maximum based on wintertime accumulation of ozone or its precursors in the Arctic. The particularly strong springtime maximum at Mauna Loa Observatory (Hawaii) is attributed to long-range transport of Asian pollution over the North Pacific in spring. A sensitivity simulation without nonmethane hydrocarbons (NMHCs) indicates small decreases of ozone concentrations (<15%) in the remote troposphere and a 20% increase in the global mean OH concentration. Without NMHCs as a source of peroxyacetylnitrate, concentrations of NOx decrease by 30% in the remote lower troposphere but increase by 70% in the continental lower troposphere and by 40% in the upper troposphere. Biogenic isoprene accounts for about half of the NMHC effects in the model.