Gametoclonal variation among anther culture-derived plants of three barley genotypes was estimated on the basis of cytological analysis (DH1, DH2 generation), observation of morphological variants (DH2, DH3) and chlorophyll mutation test (DH2, DH3). Individual head rows were grown in the field to detect possible chimeric structure of regenerants and to assess the number of variants and mutations in each line. Spontaneously doubled plants were the most frequent class (70%) among regenerants and almost 90% of them were completely fertile. There was a difference in proportion of haploids produced by different genotypes, but the highest frequency observed did not exceed 21%. The remaining regenerants were tetraploid, and contained chromosomal mutations or chimeras. In total, there were about 15% of polyploids and plants carrying chromosomal aberrations (translocations, inversions) among DH1 individuals. The changes in chromosome number and structure were the main source of observed variation. The level of gene mutation induced in vitro was relatively low. No more than 1% of microspore-derived plants expressed visible morphological changes in DH2 progeny. Only two morphological variants derived from the Bruce cultivar proved to be homozygous mutants (dwarf type) stable up the to third generation. The frequency of DH plants carrying chlorophyll mutation was 5.8%, but most of them (82%) were chimeric and had only a small mutation sector. The level of gametoclonal variation depended on the donor plant genotype. The highest proportion of variants and mutations was observed among DH plants derived from the Bruce cultivar, while the lowest was recorded among plants regenerated from anther culture of the doubled haploid line H930-36. Mechanisms leading to the observed variation and implications resulting from the presented experiments concerning implementation of anther culture in barley breeding were discussed. It was concluded that this method resulted in a high frequency of spontaneous doubling, a low frequency of genetic changes, and being less time and effort-consuming than the 'Bulbosum' technique, can be applied to most barley breeding programs.