The article analyzes models of covariance in the estimation of individual and population tree growth curves. The research follows a quantitative inferential approach, the population includes species Almendrón, ten trees per species and a series of width of rings of growth per tree. The series were smoothed by moving averages to maximize the trend due to biological growth. Results: model with random effects associated to each parameter of a second order polynomial was the most efficient in estimating the growth of species. Conclusion: random effects allow to verify the variability of the growth of a species in natural forests.