It is of interest to obtain a reliable estimate and statistical test of the simple effects of genotypes at specificlocations if a significant genotype–location interaction exists in multilocation trials commonly conductedin plant breeding and variety testing programs. For this task this study compared two different statisticalprocedures, i.e., the separate analysis of variance (ANOVA), being applied usually in the practical analysisof the multilocation trials with random locations, and the best linear unbiased prediction (BLUP) basedon a variance component mixed model, by simultaneous analysis of various cultivar multilocation trailsof rape in case of random locations. The phenotypic correlation of genotype effect prediction betweentrial locations and the average variance of differences between the genotype predictions were used,respectively for assessing the precision and the efficiency of the analysis procedures in evaluating thelocation-specific genotype effects. The result showed that the use of the separate ANOVA and the BLUPcould result in some difference in prediction, ranking and statistical testing of location-specific genotypeeffects. Compared to the separate ANOVA, the BLUP based a variance component mixed model providedboth a high precision and a high efficiency in prediction of the location-specific genotype effects for therape trials with random locations.