Field data from a population of yellow perch Perca flavescens in Saginaw Bay, Lake Huron, were used to evaluate the ability of two energetic models to predict consumption by yellow perch. Field estimates of daily ration for age-l-4 fish during May through October 1987 and 1988 were compared with independent predictions made by the Wisconsin energetic model and an energetic model developed by Karas and Thoresson. Predictions of daily ration using the Wisconsin model were lower than daily rations estimated from field data for all ages, primarily due to poor model-field agreement at temperatures above 22??C. This caused estimates of cumulative consumption from the Wisconsin model to be 25-50% lower than field estimates. Predictions of daily ration by the Karas-Thoresson model agreed with field estimates over a temperature range of 1026??C for age-1-3 yellow perch but not for older fish. Despite improvement, model predictions of cumulative consumption were 2-35% lower than field estimates. Although these tests of predicted and estimated rations may provide insight into which model produced more accurate results, it must be emphasized that field measures of daily ration are also estimates and may be in error, particularly at temperatures above 22??C where gastric evacuation rates were estimated. The Karas-Thoresson modification of the Wisconsin energetic model produced better fits to field ration data and is recommended for model applications.