Reliable methods are urgently needed for recommending minimum instream flows to protect aquatic life. For basinwide planning purposes, simple methods, which require little or nor field investigations, are required. While these methods tend to be conservative, the degree to which they protect habitat for fish is seldom determined. We applied physical microhabitat models (PHABSIM) for nine target fish species in four streams in the upper James River basin, Virginia, U.S.A., to (1) identify optimum flows to protect the fish fauna, (2) investigate the relationship between optimum flow and average discharge, and (3) compare our findings with recommendations based on simple discharge methods. Micohabitat availability for riffle-dependent species was most limited at low flows while microhabitat availability for pool-dependent species was most limited at high flows. At each study site there was a rapid increase in riffle habitat as discharge increased above zero. Optimum flow was the flow that maximized habitat for the most habitat-limited species or life stages. The recommended optimum flow increased with increased stream size but the slope was not constant; as stream size increased lower proportions of average discharge were required to maintain optimum habitat. Aquatic Base Flow recommendations (i.e. September median flow) provided varying, but reasonable degrees of habitat protection. The Montana method 10 per cent average discharge recommendation correctly identified degraded or poor habitat conditions and the 30 per cent recommendation corresponded to near optimum habitat in small streams but greater than optimum flow at the large stream site. Seven-day, 1-in-10-year low flows (7Q10) provide very limited amounts of physical habitat for riffle-dwelling fishes. The results provide a basis for making preliminary flow recommendations in this region with readily available data. Future studies will be needed to test the assumptions that need to be made.