Maps of temporal and spatial values of annual averages of daily particulate matter (PM10) concentrations were generated throughout the state of California using uncertain forms of physical data. The PM10 estimates were derived in an integrated space/time domain using the Bayesian maximum entropy (BME) mapping approach of modern spatiotemporal geostatistics. The approach possesses some interesting features which allow an insightful analysis of the PM10 space/time distribution. A complete stochastic characterization of the pollutant involves the probability density function of the PM10 map, which is the result of a rigorous knowledge-integration process. This process is considerably flexible, it can account for several physical knowledge bases and sources of uncertainty, and it may involve Bayesian or material conditionalization rules. Taking advantage of BME's flexibility, PM10 estimates were chosen which offered an appropriate representation of the real distribution in space/time, and a meaningful assessment of the representation accuracy was derived. Depending on the space scales/timescales considered, the PM10 distributions depicted considerable levels of variability, which may be associated with topographic features, climatic changes, seasonal patterns, and random fluctuations. The importance of integrating soft information available at surrounding sites as well as at the estimation points themselves was discussed. Comparisons were designed which demonstrated the usefulness of the BME-based maps to represent PM10 distributions in space/time. Areas were identified where the annual PM10 geometric mean reached or exceeded the California standard, which is valuable information for regulatory purposes. ¿ 2001 American Geophysical Union