Paloheimo and Dickie (J. Fish. Res. Bd. Canada 23: 869–908, 1966) described a regular variation between growth efficiency and ration level of fishes, termed a "K-line." Subsequent work has either confirmed the phenomenon, or suggested that it is a special case observable only at high ration levels. Two recent published experiments are reanalyzed with the aid of an analytic model based upon the metabolic requirements of fishes. The reality of the K-line is confirmed, and it is concluded that it occurs because metabolic expenditure upon spontaneous activity increases more than proportionately to increased rations, increasing the fraction of ration energy dissipated as metabolism, and thereby decreasing growth efficiency.