This study compared the variations in the mass of certain particles at an urban site, Washington, DC, and at a remote site, Shenandoah National Park, VA, in the eastern United States. Seven years (1991-1997) of Interagency Monitoring of Protected Visual Environments (IMPROVE) fine particulate matter (PM2.5), PM10, coarse fraction, SO4(2-), and total sulfur data were used for this study together with available meteorology/climatology data. Various statistical modeling and analysis procedures, including time series analysis, factor analysis, and regression modeling, were employed. Time series of the constituents were divided into four terms: the long-term mean, the intraannual perturbation, the interannual perturbation, and the synoptic perturbation. PM2.5 at the two sites made up approximately 72% of the total mass for PM10, and the coarse fraction made up the remaining 28%, on average. Thirty-one percent of the PM2.5 at the DC site and 42% at the Shenandoah site was SO4(2-), based on average data for the entire period. At the DC site, the two main contributors to the constituent mass were the long-term mean and the synoptic perturbation terms, and at the Shenandoah site, they were the long-term mean and the intra-annual perturbation terms. This suggested that the constituent mass at the two sites was affected by very different processes. The terms that provided the principal contribution to the constituent mass at the two sites were studied in detail. At the DC site, dew point trends, a climate variable, were the primary driver of the 7-year trends for PM2.5, PM10, the coarse fraction, and total sulfur, and SO2 emission trends were the primary driver of the trends for SO4(2-). SO2 emission trends influenced the trends for PM2.5 and total sulfur, appearing as the second term in the model, but only parameters dealing with climate trends had significant effects on the trends for PM10 and the coarse fraction. At the Shenandoah site, only parameters dealing with climate trends affected long-term particle trends.