An attempt to construct a mass balance of arsenic in the world atmosphere showed that the published data on arsenic concentrations in rain were not compatible with measured values of atmospheric concentrations at remote sites and with estimates of arsenic fluxes into the atmosphere. To resolve this problem, samples of rainwater and snow from eight sites in California, Washington, and Hawaii were analyzed for arsenite, asenate, and methylated forms of arsenic. The inorganic species were detectable in most samples, but no methylated forms were present above the detection limit of 0.2 ppt. Between October 1976 and March 1978, 43 samples of rain were collected at three locations near the coast in La Jolla. No significant differences between these sites were evident. The average concentration, weighted for rainfall amounts, was 0.007 ppb arsenite and 0.012 ppb arsenate giving a total concentration of 0.019 ppb As. The samples from Kauai gave an average total arsenic identical to that from La Jolla. This suggest that the La Jolla samples, most of which were collected during strong onshore flow of air from the Pacific, represent very clean air. During some periods of pollutant buildup, values up to 0.59 ppb were found in La Jolla. In a few samples, on the other hand, the arsenic concentrations were below the detection limit of 0.004 ppb. Comparable values were also found in samples of snow from Norden, California, a site at 2225 m elevation in the Sierra Nevada. These values fit well with concentrations modeled on the basis of aerosol analyses from remote sites. The average arsenic concentration at Anacortes Island, Washington, was significantly higher: 1.06 ppb with 88% of the arsenic in the form of arsenite. This value can be explained by a Gaussian plume model with the Tacoma smelter at its origin. This plant, which is 154 km from the sampling site, emits ~180 kg of arsenic per day in the form of arsenic trioxide, which is transported northward by the prevailing winds. The ratio of arsenite to arsenate in rain varies over a wide range. This is interpreted as a result of different oxidation states of arsenic in the source emissions and to redox reactions within atmospheric particulates and hydrometeors. By using the data from this study, a mass balance can be constructed for arsenic in the atmosphere which does not require biogenic methylation or hitherto unknown processes as a source for atmospheric arsenic.