A 73-day feeding experiment was conducted to evaluate the effects of dietary crude protein level on growth of Eurasian perch juveniles. Determination of the protein requirement of 2.9 g perch was performed with six semi-purified diets containing crude protein levels ranging from 0 to 60%. Fish were held in three recirculated systems, each diet being tested in triplicate. Significant differences in growth and survival appeared among fish fed diets containing up to 30% crude protein and fish fed diets from 40% crude protein and above. Three mathematical models (the four parameter saturation kinetics, broken line and second order polynomial Brett model) were used to analyse the relationships between dietary crude protein and specific growth rate. According to the saturation kinetics and polynomial models, the optimum dietary crude protein levels were 36.8 and 43.6%, respectively. Based on broken line and polynomial models, maximum growth occured at 43.1 and 56.5%, respectively. Endogenous protein losses of 2.9 g juvenile perch reached 1.11 mg fish-1 day-1 at 23°C.