The aim of this study was to evaluate the performance of seven popcorn (Zea mays L.) populations of Avati Pichinga landrace, established in different sites of Brazil and Paraguay, determining and analyzing the empirical best linear unbiased predictor (EBLUP) of population effect, and estimating covariance parameters by the restricted maximum likelihood (REML) method, for commercial and agronomic characteristics. Ten models were investigated for fitting an appropriate intergenotype covariance structure. Akaike and Bayesian information criteria were used to compare alternative models. The covariance structures chosen were Toeplitz with three bands, for grain yield; Toeplitz with two bands for both popping expansion and plant height; autoregressive for ear height; and variance component for blooming. Population effect was significant for all the agronomic traits, unlike genotype-environment interaction. Spearman correlation coefficients were not significant between grain yield and popping expansion ranking. P4 and P5 can be considered as promissory populations according to their predicted breeding values for grain yield; P1 and P4 for popping expansion. These populations can be regarded as useful for popcorn breeding programs.