Runs of homozygosity (ROH) occur in the genome as long continuous segments of homozygous genotypes that allow a reliable assessment of inbreeding levels. Today, they are widely used, and we can use them to detect the effects of autozygosity, although there is still controversy about which method of their detection is optimal. The goal of this paper was to compare three software packages with different algorithms for detection of ROH segments, namely cgaTOH, PLINK and RZooROH. For this purpose, 60 populations with different effective sizes (Ne=25 and Ne=300) were simulated through the AlphaSimR software package over 50 generations, and ROH segments with each software package were detected on them, and the results were compared with the Tukey-Kramer HSD test. Comparison results of software packages show a statistically significant difference (p < 0.01) of the mean values of small and medium segments of populations with Ne=25 obtained by the software package RZooROH versus cgaTOH and PLINK until they differ from each other. In populations with Ne=300, all three software packages are statistically significantly different (p < 0.01) in small segments, while in large segments there is no difference, or it is present in RZooROH. With its algorithm, RZooROH determines ROH segments much more restrictively and declares them identical in origin, especially for segments up to 8MB in size. For large segments and less expected autozygosity, these differences are smaller. A graphical representation of the homozygosity status of the simulated genomes versus estimation via software packages also confirms this. Further research is certainly needed that would include simulating the segregation of segments across generations and then the ability to detect those segments by software packages.