The study of the lactation curve allows the estimation of partial or total milk production during a given period. Parameters describing milk production, estimated through different linear (linear, quadratic and linear hyperbolic) and nonlinear (incomplete gamma function, exponential parabolic, Wiltmink, Papajcsik and Bordero, Cobby, and Brody) models have been studied. Data were recorded from Ojalada ewes managed under a grazing system for the production of quality label suckling lambs (Lechazo de Castilla y León). Individual records of milk production (n=6600) from 1320 single lambing ewes, between the first and the fifth week of lactation were recorded When comparing production data with production estimates using different models, the Cobby, Brody and hyperbolic linear models best estimated milk production during the first and the fifth weeks. The best estimation of the lactation peak (1509 mL) was the Papajcsik and Bordero model, between 2.39 and 3.56 weeks, with a production ranging between 1337 and 1509 mL, depending on the model. Fitting was similar among equations, obtaining similar adjusted coefficients of determination (R2Adj) and a root mean square (RMS), although nonlinear models showed greater R2Adj (0.98 except Wiltmink function with 0.88) than linear ones (0.96-0.97); the RMS was also lower in non-linear (426-485 mL) than in linear models (526-685 mL).