Although telemetered heart rate (fH) has been used as a physiological correlate to predict the metabolic rate (as oxygen consumption, V̇O2) of fish in the field, it is our contention that the method has not been validated adequately for fish. If fH in fish is to be used to estimate V̇O2, a single linear (or log-linear) relationship must be established for each species between the two variables which allows V̇O2 to be predicted accurately under all environmentally relevant conditions. Our analyses of existing data indicate that while a good linear (or log-linear) relationship can be established between fH and V̇O2, the conditions under which the relationship applies may be quite restricted. Physiological states and environmental factors affect the relationship between fH and V̇O2 significantly such that several curves can exist for a single species. In addition, there are situations in which fH and V̇O2 do not covary in a significant manner. In some situations fH can vary over much of its physiological range while V̇O2 remains constant; in others V̇O2 may vary while fH is invariate. The theoretical basis for this variability is examined to explain why the use of telemetered fH in predicting V̇O2 of fish may be limited to certain specified applications.