Nitrogen intakes and weight changes of male weanling Wistar rats fed two or more levels of test proteins or a non-protein diet for 2 weeks were determined. The slope-ratio plot for multiple-dose assays showed clearly that proteins markedly deficient in lysine yielded invalid assays. NPR values for lysine-deficient protein were markedly higher when determined at the 5% than at the 10% level of protein but foods limiting in methionine, threonine, or lysine and threonine yielded similar values at the two levels of dietary protein. It was concluded that protein quality per se is not a fixed value and depends on the level of protein fed even at low and moderate concentrations. When PER, NPR, and slope-ratio values for several samples were expressed as a percentage of the values for casein, NPR and slope-ratio methods gave almost identical results but PER yielded low values for poor quality proteins. However, the three methods rated the eight test proteins in the same relative order.