The occurrence of arsenic in groundwater is a recognized environmental hazard with worldwide importance and much effort has been focused on surveying and predicting where arsenic occurs. Temporal variability is one aspect of this environmental hazard that has until recently received less attention than other aspects. For this study, we analyzed 1245 wells with two samples per well. We suggest that temporal variability, often reported as affecting very few wells, is perhaps a larger issue than it appears and has been masked by datasets with large numbers of non-detect data. Although there was only a slight difference in arsenic concentration variability among samples from public and private wells (p = 0.0452), the range of variability was larger for public than for private wells. Further, we relate the variability we see to geochemical factors—primarily variability in redox—but also variability in pH and major-ion chemistry. We also show that in New England there is a weak but statistically significant indication that seasonality may have an effect on concentrations, whereby concentrations in the first two quarters of the year (January–June) are significantly lower than in the second two quarters (July–December) (p < 0.0001). In the Central Valley of California, though not statistically significant (p = 0.4169), arsenic concentration is lower in the first quarter of the year but increases in subsequent quarters. In both regions, these changes appear to follow groundwater levels. It is possible that this difference in arsenic concentrations is related to groundwater level changes, pumping stresses, evapotranspiration effects, or perhaps mixing of more oxidizing, lower pH recharge water in wetter months. Focusing on the understanding the geochemical conditions in aquifers where arsenic concentrations are concerns and causes of geochemical changes in the groundwater environment may lead to a better understanding of where and by how much arsenic will vary over time.