Snow samples were collected in southeastern Idaho over two winters to assess trace element and common ion concentrations in air pollutant fallout across the region. The objectives were to: (1) develop snow sampling and analysis techniques that produce accurate and ultra-low measurements of a broad suite of fallout elements, (2) identify the spatial and temporal trends of the fallout elements across the region, (3) determine if there are unique combinations of fallout elements that are characteristic to the major source areas in the region (source area profiles), and (4) use pattern recognition and multivariate statistical techniques (principal component analysis and classical least squares regression) to investigate source area apportionment of fallout concentrations measured at downwind locations where plumes from different source areas might mix. In the winter of 2000-2001, 250 snow samples were collected across the region over a 4-month period and analyzed in triplicate using inductively coupled plasma mass spectrometry (ICP MS) and ion chromatography (IC). Thirty nine trace elements and nine common ions were positively identified in most samples. The data were analyzed using pattern recognition tools in the software, Pirouette((R)) (Infometrix). These results showed a large crustal component (Al, Zn, Mn, Ba, and rare earth elements), an overwhelming contribution from phosphate processing facilities located outside Pocatello in the southern portion of the Eastern Snake River Plain, some changes in concentrations over time, and no obvious source area profiles (unique chemical signatures) other than at Pocatello. Concentrations near a major U.S. Department of Energy industrial complex oil the Idaho National Engineering and Environmental Laboratory (INEEL) were lower than those observed at major downwind communities. In the winter of 2001-2002, a new sampling design was tested and 135 additional samples collected to estimate pure emission profiles from the major source areas in the region. Classical least squares regression (CLS) was then used to source apportion these profiles at downwind mixing sites where plumes from the different source areas mixed. CLS performed reasonably well, predicting 36-58% of the total fallout concentrations measured at the mixing sites. (C) 2004 Elsevier B.V All rights reserved.