We have applied a global three-dimensional chemistry-transport model to quantify the photochemistry of tropospheric O-3 and compare the main source categories. We simulated a 15 year period (1979-1993) on the basis of the European Centre for Medium-Range Weather Forecasts meteorological reanalyses and a time-varying emission data set. We calculate that stratosphere-troposphere exchange (STE) strongly contributes to O-3 in regions where the photochemistry is quiescent. Since such regions play a minor role in radiative and chemical processes, we argue that STE-derived O-3 is much less important than is suggested by its column abundance. By distinguishing between photochemical pathways in the model we calculate that tropospheric O-3 in the extratropical Northern Hemisphere is strongly affected by industrial and fossil fuel-related emissions. In the tropics and Southern Hemisphere, natural emissions still play a major role. Our model results indicate a less important role for man-made biomass burning emissions than previous analyses. Further, the results show that tropospheric O-3 trends are strongly influenced by transports of pollution and by meteorological variability. Scenario calculations for the year 2025 suggest that man-made emissions at low northern latitudes, in particular in southern and eastern Asia, will become a very strong tropospheric O-3 source in the next decades. This will influence O-3 levels on a hemispheric scale so that despite pollution regulations in Europe and North America, surface O-3 will continue to grow.