We present procedures for classifying stream reaches based on their streamflow regime and predicting these hydrologic classes from watershed attributes. We first identified a set of twelve streamflow variables relevant to biota and computed them for 543 minimally impacted gauged watersheds in the Western USA.  We then used Principal Component Analysis and K-means clustering to classify the reaches into 8 different hydrologic groups. Linear Discriminant Models (LDM), Classification and Regression Trees (CART), and Random Forests (RF) models were then developed to predict hydrologic class from watershed attributes. Our classification differs from previous hydrologic classifications in that magnitude of discharge is used to discriminate small streams from big streams.  This difference should better relate hydrologic classes to stream biota given the strong biotic gradients that occur with stream size. We found that the RF model gave a slightly better prediction of stream class than the other models evaluated (% correct predictions = 77 for RF, 57 for CART and 73 for LDM).  Prediction of the hydrologic class of ungauged reaches of stream should allow better prediction of the biotic composition expected in a given reach and may aid in understanding when biotic impairment is caused by hydrologic alteration.