Habitat quality and quantity strongly influence the viability and abundance of stream fish populations and are a major focus of monitoring and assessment programs. While our ability to measure habitat quantity has rapidly improved with new analytical methods, the measurement of habitat quality remains challenging. A promising development on this front has been the application of drift-foraging bioenergetics models to predict Net Rate of Energy Intake (NREI; energy gains - losses) as a function of habitat attributes (e.g., depth and velocity) and prey abundance. Because NREI models mechanistically link habitat to energy gain, an ecologically meaningful currency of fitness, they have been increasingly used as tools to assess habitat quality for management purposes. However, simplifying the drift-foraging process for modelling can introduce process errors that complicate model interpretation. These include: (1) a mismatch in timescales among sub-models that affect how NREI varies with temperature; and (2) uncertainty in how temporal variation in foraging strategy influences model predictions. These issues are especially problematic when foraging windows of fish are restricted due to predation risk, which we illustrate with an empirical case study of the diel foraging patterns of rainbow trout in the Skagit River, BC. We discuss the implications of these issues for interpreting habitat quality estimates from drift-foraging models and offer some potential solutions.