Abstract. The aim of the study was to determine the genetic variability of the Polish sheep breeds Świniarka, Wrzosówka, Pomorska, and Wielkopolska based on mitochondrial control region polymorphism. A comprehensive phylogenetic analysis and information about the genetic origin of the breeds were also obtained. The genetic variability of the breeds studied has been assessed based on the number of haplotypes, haplotype diversity, nucleotide diversity, the average number of nucleotide differences, the number of mutations, and phylogenetically informative sites. Sequence divergence between identified haplogroup A (HA) and haplogroup B (HB) was also calculated. Moreover, a neighbour-joining (NJ) haplotype tree was constructed based on Kimura's two-parameter genetic distance calculation. Finally, the history of the population was investigated by mismatch distribution and Fu's F statistics. The 559 bp long mitochondrial DNA (mtDNA) control region (CR) sequences of 143 sheep were analysed. The 65 haplotypes were defined by 45 parsimony informative sites. Among the four Polish breeds, Wrzosówka had the highest while Świniarka the lowest values of haplotype (H_d) and sequence diversity (π) (H_d =  0.9735 and π =  0.0040 for Wrzosówka; H_d =  0.8975 and π =  0.0030 for Świniarka). Five haplotypes were shared between breeds, whereas the remaining 60 were unique. The NJ phylogenetic tree has revealed that 61 haplotypes of all analysed breeds clustered into clade B while the remaining 4 haplotypes representing all but the Świniarka breed pooled together with clade A. None of the other reported mitochondrial haplogroups were identified. The haplotypes representing HB formed a star-like network with the single central haplotype, which in association with extensive haplotype sharing reveals a weak structure of Polish breeds and the existence of gene flow between the breeds studied.