Non-invasive DNA genotyping using hair samples has become a common method in the population survey of Asiatic black bears (Ursus thibetanus) in Japan. However, the accuracy of the genotype data was rarely discussed in empirical studies. Therefore, we conducted a large-scale pilot study to examine the genotyping accuracy and sought the efficient way of error-checking of hair-trapping data. The presence of erroneous genotypes among the identified individuals was checked by a post-hoc goodness-of-fit test that determined the match between the expected and observed frequencies of individual homozygote at 0 to 6 loci. The result indicated that the samples with less than 10 hairs have a high possibility of erroneous genotypes such as allelic dropout. Therefore, for better accuracy, it is recommend for using only >10 hairs samples in genotyping and running the post-hoc goodness-of-fit test to exclude erroneous genotypes before taking the data to downstream analysis such as capture-mark-recapture estimation.