A total of 123 water samples were examined in parallel by culture and semi-nested PCR for the presence of Legionella. They comprised: 35 samples of hot water distributed by the urban municipal water supply system (MWSS) taken in institutions, 45 samples of hot water distributed by urban MWSS taken in dwellings, 27 samples of cold water distributed by rural MWSS taken in dwellings, and 16 samples of cold well water taken in rural areas. The greatest frequency of the isolation of Legionella by culture (88.6%) was recorded in the samples of hot water from the urban institutions, having been greater compared to all other sources (p<0.001). The frequency of Legionella isolation from hot water in urban dwellings (28.9%) was significantly greater compared to the combined value (2.3%) for cold water from rural MWSS and wells (p<0.001). Strains belonging to Legionella pneumophila serogroups 2-14 predominated in the examined samples, while strains of L. pneumophila serogroup 1 and strains of Legionella spp. (other than L. pneumophila) were 3-fold less numerous. The rates of positive findings in the semi-nested PCR (stage 2) were greater than culture isolations in all kinds of samples, except for urban institutions. The correlation between the culture and PCR results was positive for samples of hot water from urban MWSS (p<0.01), but not for samples of cold water from rural MWSS and wells (p>0.5). A significant correlation was found between rates of PCR-positive results and numbers of Legionella pneumophila serogroups 2-14 strains, but not for other Legionella serogroups or species. In conclusion, our results support the opinion that though PCR cannot be a substitute for the isolation of Legionella by culture, it could be regarded as an useful complementary method.