
Adoption of mixed models that estimate genetic parameters and predicted genotypic values correctly are important in increasing the efficiency of breeding programs, as decision-making during the selection process. The aim of the present study was to estimate variance components via REML and predict genotypic values of simple maize hybrids via BLUP. Trial was conducted in the city of Frederico Westphalen, RS, Brazil. Pre-commercial hybrids used were coming from the KSP Seeds Ltd. company's breeding program, located in the city of Pato Branco-PR. A randomized complete block design with three replications was used. Twenty-four pre-commercial simple hybrids (KSP Seeds Ltd..) were used in trial's conducting. Assessed traits were: 1. ear insertion height (EH) 2. Plant height (PH) 3. ear diameter (ED) 4. ear length (EL): 5. Average number of rows (NR) 6. number of kernels per row (NKR) 7. grain yield (GY) 8. Hundred-kernels weight (HKW). Deviance analysis was performed with and without heritability estimates thus generating obtaining their deviances, by subtracting the reduced model of the complete model it was obtained the likelihood ratio test (LRT), compared with tabulated value of the chi-square test one degree of freedom. Estimates of genetic parameters were obtained by restricted Maximum Likelihood and Best Linear Unbiased Prediction models (REML / BLUP), using Selegen software (Resende, 2007). Presence of genetic variation among the studied hybrids indicate that can have genetic gain in these maize hybrid's breeding program. KSP22 pre-commercial simple maize hybrid has high performance for grain yield, enabling their inclusion in regional trials with commercial checks.