Real-time and integrated measurements of gaseous and particulate pollutants were conducted inside five conventional diesel school buses, a diesel bus with a particulate trap, and a bus powered by compressed natural gas (CNG) to determine the range of children's exposures during school bus commutes and conditions leading to high exposures. Measurements were made during 24 morning and afternoon commutes on two Los Angeles Unified School District bus routes from South to West Los Angeles, with seven additional runs on a rural/suburban route, and three runs to test the effect of window position. For these commutes, the mean concentrations of diesel vehicle-related pollutants ranged from 0.9 to 19 microg/m(3) for black carbon, 23 to 400 ng/m(3) for particle-bound polycyclic aromatic hydrocarbon (PB-PAH), and 64 to 220 microg/m(3) for NO(2). Concentrations of benzene and formaldehyde ranged from 0.1 to 11 microg/m(3) and 0.3 to 5 microg/m(3), respectively. The highest real-time concentrations of black carbon, PB-PAH and NO(2) inside the buses were 52 microg/m(3), 2000 ng/m(3), and 370 microg/m(3), respectively. These pollutants were significantly higher inside conventional diesel buses compared to the CNG bus, although formaldehyde concentrations were higher inside the CNG bus. Mean black carbon, PB-PAH, benzene and formaldehyde concentrations were higher when the windows were closed, compared with partially open, in part, due to intrusion of the bus's own exhaust into the bus cabin, as demonstrated through the use of a tracer gas added to each bus's exhaust. These same pollutants tended to be higher on urban routes compared to the rural/suburban route, and substantially higher inside the bus cabins compared to ambient measurements. Mean concentrations of pollutants with substantial secondary formation, such as PM(2.5), showed smaller differences between open and closed window conditions and between bus routes. Type of bus, traffic congestion levels, and encounters with other diesel vehicles contributed to high exposure variability between runs.