The Kathmandu Valley is a bowl-shaped basin in the Nepal Himalaya, with a rapidly growing city surrounded by rice fields and steep terraced and forested mountain slopes. The valley's air quality is influenced by urban and rural emissions, nocturnal pooling of cold air, slope winds, and a daily exchange of air through mountain passes. To understand these processes and to inform air pollution policy in Nepal, we have carried out the most comprehensive study of air pollution in Nepal to date. During the 9-month dry season of 2004-2005, we carried out continuous measurements every minute of carbon monoxide, ozone, PM10, wind speed, wind direction, solar radiation, temperature, and humidity on the eastern edge of Kathmandu city, at a site that daily received air from both the city and rural areas. We recorded the diurnal cycle of the vertical temperature structure and stability with temperature loggers on towers and mountains. A sodar measured the mixed layer height and upper-level winds. 24-hour simultaneous bag sampling campaigns on mountain peaks, passes, the rural valley, and within the city provided glimpses of the spatial patterns of the diurnal cycle of CO -- a useful tracer of anthropogenic emissions. We measured winds on mountain passes and ozone on mountain peaks. At our main measurement site we found a daily-recurring pattern of CO and PM10, with an afternoon low showing rural background levels, even though the arriving air had traversed the city. This was followed by an evening peak starting at sunset, a second low late at night, and a morning peak enhanced by re-circulation. Pollutants emitted in the valley only traveled out of the valley between the late morning and sunset. During winter months, rush hour was outside of this period, enhancing the morning and evening peaks. Within the city, ozone dropped to zero at night. At mid-day we observed an ozone peak enhanced by photochemical production when the air mass that had been stagnant over the city swept past our site at the edge of the city. At nights on surrounding mountains, ozone remained high, while CO dropped to regional background levels. Using the MM5 model, we simulated the Kathmandu Valley's meteorology for selected days in February and May 2005. The model successfully captured essential features of the valley's meteorology. We used it to understand the origin and pathways of air arriving at our sites, and to address specific questions. At night polluted air was found to be lifted off the ground but not removed from the valley; it was mixed down again in the morning. The break-up of the valley's nocturnal temperature inversion was dominated in February by up- slope flows along the valley rim mountains, plus subsidence over the fog-covered valley center. In May, when there was no fog, surface heating of the valley floor also contributed. During the mornings some pollutants were ventilated out of the valley by upslope flows. From late mornings through afternoons, strong westerly winds arriving through the valley's western passes brought rural background air into the city, while rapidly sweeping the night and morning's emissions out the eastern passes. Pollutants emitted into the afternoon westerly winds left the valley in less than 2 hours; pollutants emitted after the winds ceased at sunset lingered for up to 18 hours. The timing of emissions therefore has a bigger effect on air quality within the valley than the daily total emissions; this is important for policy considerations.