Six mathematical models were evaluated to predict nitrogen (N) optimum rate for the potato production in an experiment carried out in a Red Yellow Cambic Podzolic soil. Five N rates (0; 50; 100; 200 and 300 kg ha-1) were evaluated in a completely randomized block design, with four replications. N fertilizer was applied in furrows, as ammonium sulphate. Monalisa cultivar was planted on May 5 following the recommended agronomic practices for the potato. Tubers were harvested and weighed on September 14. Six models, linear plateau, quadratic plateau, Mitscherlich, sigmoidal, square root and quadratic, were selected and evaluated based upon the criteria: biological logic, regression mean square residue, significance of F-test or lack of adjustment, high determination coefficient value (R²), significance of the regression parameters, pattern absence in the graph of residues versus foreseen values and, when possible the money left over after selling the potato and paying the N fertilizer. For the adopted criteria, the quadratic model was chosen and the economic N rate was 163 or 171 kg ha-1, at unfavorable or favorable potato price situations, respectively.
Solanum tuberosum L.; nitrogen; economic rate; fertilizer