Single nucleotide polymorphisms (SNPs) are replacing microsatellites for population genetic analyses, but it is not apparent how many SNPs are needed or how well SNPs correlate with microsatellites. We used data from the gopher tortoise, Gopherus polyphemus – a species with small populations, to compare SNPs and microsatellites to estimate population genetic parameters. Specifically, we compared one SNP dataset (16 tortoises from 4 populations sequenced at 17,901 SNPs) to two microsatellite datasets, a full dataset of 101 tortoises and a partial dataset of 16 tortoises previously genotyped at 10 microsatellites. For the full microsatellite dataset, observed heterozygosity, expected heterozygosity, and FST were correlated between SNPs and microsatellites; however, allelic richness was not. The same was true for the partial microsatellite dataset, except that allelic richness, but not observed heterozygosity, was correlated. The number of clusters estimated by Structure differed for each dataset (SNPs = 2; partial microsatellite = 3; full microsatellite = 4). PCA showed four clusters for all datasets. More than 800 SNPs were needed to correlate with allelic richness, observed heterozygosity, and expected heterozygosity, but only 100 were needed for FST. The number of SNPs typically obtained from NGS far exceeds the number needed to correlate with microsatellite parameter estimates. Our study illustrates that diversity, FST, and PCA results from microsatellites can mirror those obtained with SNPs. These results may be generally applicable to small populations, a defining feature of endangered and threatened species, because theory predicts that genetic drift will tend to outweigh selection in small populations.