Although single-step GBLUP (ssGBLUP) is a breeding value method, single-nucleotide polymorphism (SNP) effects can be backsolved from ssGBLUP genomic estimated breeding values (GEBV), and p-values can be obtained as a measure of estimation certainty. This enables single-step genome-wide association studies (ssGWAS). However, obtaining p-values for ssGWAS relies on the inversion of dense matrices, which poses computational limitations in large genotyped populations. In this study, we present an algorithm to approximate p-values for SNP in ssGWAS with many genotyped animals. The approximation relies on the algorithm for proven and young (APY) and submatrices for core animals. To test that, we first compared SNP p-values obtained with an exact inversion using the genomic relationship matrix (G-1) for 50K genotyped animals to those estimated with an exact inversion using GAPY-1and those obtained with the proposed approximation based on GAPY-1. Then, we compared these results with those obtained with the proposed approximation using 450K genotyped animals. The same genomic regions in chromosomes 7 and 20 were identified with p-values obtained with G-1, GAPY-1, and the approximation based on GAPY-1when using 50k genotyped animals and 1.5M in the pedigree. In terms of computational requirements, obtaining p-values with the approximation based on GAPY-1represented a reduction of 38 times in wall-clock time and ten times in memory requirement compared to using the exact inversion with GAPY-1. When the approximation was applied to a population of 450K genotyped animals and 1.8 in the pedigree, apart from the two genomic regions in chromosomes 7 and 20 previously identified with the smaller genotyped population, two new significant regions on chromosomes 6 and 14 were uncovered, indicating an increase in GWAS detection power when including more genotypes in the analyses. The process of obtaining p-value with the approximation and 450K genotyped individuals took 24.5 wall-clock hours and 87.66 GB of memory, which is expected to increase linearly with the addition of noncore genotyped individuals. With an algorithm that approximates the prediction error variance of SNP effects based on APY, ssGWAS with p-values for SNP is possible in large genotyped populations. The computational cost of obtaining p-values in ssGWAS is no longer a limitation in extensive populations with many genotyped animals.