We analyzed seasonal water samples from the Cheat and Tygart Valley river basins, West Virginia, USA, in an attempt to classify streams based on water chemistry in this coal-mining region. We also examined temporal variability among water samples. Principal component analysis identified two important dimensions of variation in water chemistry. This variation was determined largely by mining-related factors (elevated metals, sulfates, and conductivity) and an alkalinity-hardness gradient. Cluster analysis grouped water samples into six types that we described as reference, soft, hard, transitional, moderate acid mine drainage, and severe acid mine drainage. These types were statistically distinguishable in multidimensional space. Classification tree analysis confirmed that chemical constituents related to acid mine drainage and acid rain distinguished these six groups. Hard, soft, and severe acid mine drainage type streams were temporally constant compared to streams identified as reference, transitional, and moderate acid mine drainage type, which had a greater tendency to shift to a different water type between seasons. Our research is the first to establish a statistically supported stream classification system in mined watersheds. The results suggest that human-related stressors superimposed on geology are responsible for producing distinct water quality types in this region as opposed to more continuous variation in chemistry that would be expected in an unimpacted setting. These findings provide a basis for simplifying stream monitoring efforts, developing generalized remediation strategies, and identifying specific remediation priorities in mined Appalachian watersheds.