A global three-dimensional model of tropospheric chemistry is used to investigate the changes in tropospheric O3 and OH since preindustrial times as a result of fuel combustion and industry, biomass burning, and growth in atmospheric CH4. Model results indicate a 63% increase of the global tropospheric O3 burden from preindustrial times to present (80% and 50% in the northern and southern hemispheres, respectively). Anthropogenic emissions of NOx and of CO and hydrocarbons make comparable contributions to the global O3 increase (60% and 40% respectively), even though the local rate of tropospheric O3 production is generally NOx limited. The rise in O3 production parallels closely the rise in the emissions of CO and hydrocarbon because the O3 yield per mole of CO or hydrocarbon oxidized has remained constant at 0.7-0.8 mol/mol since preindustrial times. In contrast, the O3 production efficiency per mole of NOx emitted has decreased globally by a factor of 2. We find a 9% decrease in the global mean OH concentration (mass-weighted) since preindustrial times. A linear relationship is found in the model between the global mean OH concentration and the SN/SC3/2 ratio, where SN and SC are the sources of NOx and of CO and hydrocarbons, respectively. The relative constancy of the global mean OH concentration since preindustrial times reflects the conservation of the SN/SC3/2 ratio despite large increases in both SN and SC. Comparisons of model results with reconstructed nineteenth century observations of O3 at continental sites indicate a systematic overestimate of about 5 ppbv. Correcting this overestimate would require either a large missing chemical sink for O3 or a downward revision of the natural NOx source from lightning (3 Tg N yr-1 in our model). The nineteenth century observations of O3 over France show no vertical gradient between the boundary layer and the free troposphere, which is inconsistent with our current understanding of tropospheric O3. The model underestimates preindustrial CO concentrations derived from polar ice cores; these measurements are difficult to reconcile with any reasonable CO emission inventories. ¿ 1998 American Geophysical Union