The short-range atmospheric dispersion model UK-ADMS has been applied to calculate ground-level concentrations of arsenic within 20 km of the Nov¿ky Power Plant, Nitra Valley, Slovakia, as a function of distance and direction from source and year of operation, for application to an assessment of local population exposure to arsenic, and hence to two epidemiological studies of the effect of arsenic exposure on skin, lung, and bladder cancer in an east European population. A simple approach, where the effect of complex terrain is inclued by using local observations for meteorological input to the model, is sufficient for initial screening to compare atmospheric dispersion with other exposure pathways. This shows that isopleths of ground-level annual average arsenic concentration in air (excluding periods when there is no wind) are elongated aligned with the axis of the valley. The maxima of concentration occur about 2 km to the northeast and southwest of the source, and a factor of 30 reduction in concentration is found from there to a distance 10 km away. The value of this information for epidemiological studies is discussed. If atmospheric concentration is found to be an important determinant of arsenic exposure for the Nitra Valley population, it will be necessary to carry out further investigation of the dispersion of the plume in complex terrain when there is no wind in the valley, which is the case in 13% of the observations and is estimated could contribute more than 50% of the annual average arsenic concentration close to the source. ¿ 2001 American Geophysical Union