The analysis of human leukocyte antigen (HLA) class I allospecificity expression in malignant lesions has been hampered by the limited availability of HLA class I allospecificity-specific monoclonal antibodies (mAbs) which stain tissues in immunohistochemical (IHC) reactions. During the 12th International Histocompatibility Workshop, the HLA and cancer component made available a panel of mAbs capable of detecting monomorphic, locus- and allo-specific HLA class I antigenic determinants in surgically removed frozen tissue sections by IHC staining. In the present study, we have utilized this panel of mAbs to analyze the expression of HLA class I allospecificities in 33 primary and in 11 metastatic lesions surgically removed from HLA-typed patients with malignant melanoma, as this information contributes to determine the extent of HLA class I antigen abnormalities in melanoma lesions. HLA class I antigens were downregulated in six (18.2%) of the primary lesions and in six (54.5%) of the metastatic lesions. Selective loss of HLA-A and HLA-B antigens was detected in two (6.1%) and in one (3.0%), respectively, of the primary lesions, but in none of the metastases. HLA-A and HLA-B antigens were downregulated in three (9.1%) and four (36.4%) of the primary and metastatic lesions, respectively. Selective loss of one or more HLA class I allospecificities was found in 10 (33.0%) and two (18.0%) of the 33 primary and 11 metastatic melanoma lesions analyzed, respectively. HLA class I antigen abnormalities were present in 16 (48.5%) of the 33 primary lesions analyzed (i.e. six lesions demonstrating abnormal reactivity with HLA class I monomorphic-specific mAb, two lesions demonstrating selective abnormal reactivity with HLA-B locus-specific mAb, one lesion demonstrating selective abnormal reactivity with HLA-A and HLA-B locus-specific mAbs, and seven lesions demonstrating selective abnormal reactivity with HLA class I allele-specific mAb). Furthermore, HLA class I antigen abnormalities were present in nine (81.8%) of the 11 metastatic lesions analyzed (i.e. six lesions demonstrating abnormal reactivity with HLA class I monomorphic-specific mAb, one lesion demonstrating selective abnormal reactivity with HLA-A locus-specific mAb, and two lesions demonstrating selective abnormal reactivity with HLA class I allele-specific mAb). It cannot be ruled out that the frequency of HLA class I allospecificity abnormalities is higher, as the expression of several HLA class I allospecificities could not be investigated because of the lack of appropriate probes. The frequency of HLA class I antigen defects in primary lesions was significantly correlated with primary lesion thickness, an important prognostic marker in melanoma, arguing for a potential clinical significance of HLA class I antigen abnormalities in melanoma. In conclusion, the results of the present study (i) demonstrate that the frequency of HLA class I allospecificity abnormalities in primary melanoma lesions is markedly higher than that of total HLA class I antigen downregulation described in the literature; (ii) corroborate our previous findings that staining of melanoma lesions with mAb to monomorphic determinants of HLA class I antigens does not detect selective HLA class I allospecificity loss; and (iii) demonstrate for the first time selective loss of antigenic determinants expressed on HLA class I molecules in melanoma lesions. The latter finding indicates that at least two mAbs recognizing distinct antigenic determinants on the HLA molecule being investigated should be used for IHC staining of tissue sections in order to prove that lack of immunostaining reflects actual loss of the corresponding HLA molecule and not selective loss of antigenic determinants.