Radon (222Rn) concentrations in 26 ground water wells of two distinct lithologies in the Piedmont of Virginia were measured to assess variation in ground water radon concentrations (GWRC), to evaluate differences in concentrations related to well characteristics, lithology, and spatial distributions, and to assess the feasibility of predicting GWRC. Wells were sampled in accordance with American Public Health Association Method 7500 Rn-B, with modifications to include a well shaft profile analysis that determined the minimum purge time sufficient to remove the equivalent of one column of water from each well. Statistically significant differences in GWRC were found in the Trssu (1482 +/- 1711 pCi/L) and Mpg (7750 +/- 5188 pCi/L) lithologies, however, no significant differences were found among GWRC at each well over time. Using multiple regression, 86% of the variability (R2) in the GWRC was explained by the lithology, latitudinal class, and water table elevation of the wells. The GWRC in a majority of the wells studied exceed US Environmental Protection Agency designated maximum contaminant level and AMCL. Results support modifications to sampling procedures and indicate that, in previous studies, variations in GWRC concentrations over time may have been due in part to differences in sampling procedures and not in source water.