In the upper Midwest, USA, elevated arsenic concentrations in ground water are related to the presence of northwest provenance Late Wisconsinan drift. Within the footprint of this sediment, 1,690 of 7,101 (23.8%) sample locations exceed 10 ?g/l arsenic compared to only 186 of 4,333 (4.3%) outside the footprint. Wells screened in glacial drift and shallow bedrock wells overlain by this glacial sediment are most impacted by arsenic contamination. Elevated arsenic concentrations are spatially related to one another over significant distances – more than 150 kilometers. Elevated arsenic concentrations are not randomly distributed. Evidence suggests that the distinct physical characteristics of the northwest provenance Late Wisconsinan drift cause the geochemical conditions necessary to mobilize arsenic. This fine-grained, comparatively organic-rich, biologically-active sediment creates a geochemical environment that is favorable to a regional-scale mobilization of arsenic in ground water via reductive mobilization mechanisms such as reductive dissolution and reductive desorption. Arsenic concentrations are positively correlated to analytes like iron and manganese, and it is negatively correlated to analytes like chloride and bromide. In this study, there is no correlation between arsenic and carbonate, sulfate, total organic carbon, or ammonium. There is a weak negative correlation between phosphorous and arsenic. Arsenic present in ground water is aqueous, and most of the arsenic present in Minnesota ground water is As(III). No extremely low pH waters were encountered. Sediment and ground water arsenic concentrations from the same sample location are not correlated. In this study, measured total sediment arsenic concentrations are not particularly high, 0.6 mg/kg to 10.6 mg/kg. A small but significant concentration of sediment arsenic is adsorbed onto or coprecipitated with iron hydroxides; this arsenic is labile arsenic. This study highlights an important and often unrecognized phenomenon: high-arsenic sediment is not necessary to cause arsenic-impacted ground water – when 'impacted' is now defined as greater than 10 ?g/l. Arsenic contamination is more common in domestic wells and in monitoring wells than in public water system wells. Additionally, arsenic contamination is more prevalent in domestic wells with a short screen set in proximity to an upper confining unit, such as till. The geochemical environment at the interface between the confining unit and the aquifer is conducive to arsenic mobilization. Public water system wells have distinctly different well construction practices and well characteristics when compared to domestic and monitoring wells. Public water system well construction practices such as seeking a thick, coarse aquifer and installing a long well screen, yield good water quantity. These well construction practices also, coincidentally, often yield low arsenic. Changes in routine domestic well construction practices to exploit deeper parts of an aquifer and use a longer screen could reduce the number of domestic wells that are affected by arsenic contamination in the upper Midwest. Arsenic concentrations do not change systematically in newly-constructed, glacial sediment domestic wells in Minnesota. Over the course of nine months, there was no discernible pattern of arsenic concentration variability in three public water system wells. Arsenic concentration standard deviations ranged from 0.5 ?g/l to 3.5 ?g/l for the three public water system wells. Understanding and quantifying arsenic concentration variability may be of critical importance to public water systems with an average arsenic concentration just over 10 ?g/l. There is a pattern to arsenic concentration variability in some public water system wells because of reductive arsenic mobilization mechanisms. Four of eleven sampled public water system wells in Minnesota had notable arsenic concentration variability over a short period of time. In these wells, the arsenic concentration was less than 10 ?g/l shortly after pumping started, but the arsenic concentration increased over time to a level exceeding 10 ?g/l. In these wells, the arsenic concentration decreased to below 10 ?g/l again four hours after pumping stopped. The arsenic concentration variability can be explained by reductive arsenic mobilization mechanisms. The pe in three of these four wells had initial pe readings that were positive or only slightly negative, and the pe moved through the crucial positive-to-negative range, where the speciation of arsenic changes from being dominated by As(V) species to being dominated by As(III) species. Six of the seven wells that did not exhibit short-term temporal changes in arsenic concentration did not have pe measurements that traveled through the positive-to-negative pe range. Three of the four wells with arsenic concentration variability used turbine pumps. All of the remaining wells used submersible pumps. Results of this research indicate that arsenic mobilization mechanisms such as adsorption/desorption to iron hydroxides, reductive dissolution of iron hydroxides, and microbially-mediated release of arsenic are likely to play an important role in upper Midwestern ground water. Several arsenic mobilization mechanisms were either refuted or not supported based on the results of this research. These mechanisms include anthropogenic sources, anion competition for adsorption/desorption sites to various metal hydroxides, aging of iron hydroxides, resulting in a decreased number of adsorption sites, competition with or complexation with various organic species, and release from or coprecipitation with sulfide minerals (e.g. pyrite). Most of the data used in this analysis were collected for other purposes; therefore, the cost of conducting this analysis was relatively low. In spite of the low cost, the results have considerable consequence. A regional-scale environmental problem was identified, and significant knowledge gaps that inhibit planning, such as the lack of information tying sampling results to a specific well, were exposed. An important result of this analysis is demonstration of the value of finding and using existing sampling results to cost-effectively observe and characterize regional-scale environmental problems. Regional-scale problems cannot be observed with only local-scale data analysis.