Recurrent selection is a method for developing new popcorn (Zea mays L.) cultivars.We aimed to determine the selection accuracy and genetic gains for different selection strategies: estimates based exclusively on phenotypic data (PhEN), estimates based on phenotypic and genotypic data (PhEN + GEN), and estimates based exclusively on single nucleotide polymorphism (SNP) marker genotyping (GEN). For the GEN strategy, we tested, via simulation, the possibility of reducing the number of SNPs and increasing the training population. The traits evaluated were 100-grain weight, ear height, grain yield, popping expansion, plant height, and popping volume. Field trials were undertaken with 98 S 1 progenies at two locations in an incomplete block design with three replications. The progenies' parents were genotyped with a panel of ∼10,507 SNPs. As predicted by the GEN strategy at different selection intensities, the average annual genetic gain for the different traits were 29.1 and 25.2% higher than those of PhEN and GEN + PhEN for 98 candidates; 148.3 and 140.9% higher for 500; and 187.9 and 179.4% higher for 1,000 selection candidates, respectively. Recurrent genomic selection may result in high genetic gain, provided that: (a) phenotyping is accurate; (b) selection intensity is explored by genotyping several progenies and increasing the number of candidates; (c) genomic selection is used for early selection;and (d) the model is adjusted for a few more cycles of phenotyping. The simulation suggests that desirable values of genetic gain may be obtained by reducing the number of SNPs and increasing the training population size.