The Electric Power Research Institute (EPRI) is completing a review of methods used or useful for recommending instream flows below reservoirs. The three principal conclusions: no consensus exists on the number or identity of input variables needed to model the biological effects of flow alteration; although mechanistic models predicting the biological effects of flow alterations would be highly desirable, none yet exists; and, the only models that seem geographically transportable are ones that select site-specific input variables, use site-specific data transformation schemes, or adjust parameters to correspond to regional characteristics. The existing methods are equally represented by those recommending a particular flow regime from only basin or hydraulic variables without clear justification; those producing a relationship between flow and an unmeasurable habitat index (such as the U.S. Fish and Wildlife Service Habitat Suitability Indexes or Weighted Usable Area); and those producing an empirically descriptive (but not usually predictive) equation relating flow to standing crop of fish, invertebrates, or riparian vegetation. This paper discusses the results of the EPRI study and suggests future approaches to making instream flow recommendations.