Prior studies using polyclonal populations of natural killer (NK) cells have revealed that expression of certain major histocompatibility complex (MHC) class I molecules on the membrane of normal and transformed hematopoietic target cells can prevent NK cell-mediated cytotoxicity. However, the extent of clonal heterogeneity within the NK cell population and the effect of self versus non-self MHC alleles has not been clearly established. In the present study, we have generated more than 200 independently derived human NK cell clones from four individuals of known human histocompatibility leukocyte antigens (HLA) type. NK clones were analyzed for cytolytic activity against MHC class I-deficient Epstein Barr virus (EBV) transformed B lymphoblastoid cell lines (B-LCL) stably transfected with several HLA-A, -B, or -C genes representing either self or non-self alleles. All NK clones killed the prototypic HLA-negative erythroleukemia K562 and most lysed the MHC class I-deficient C1R and 721.221 B-LCL. Analysis of the panel of HLA-A, -B, and -C transfectants supported the following general conclusions. (a) Whereas recent studies have suggested that HLA-C antigens may be preferentially recognized by NK cells, our findings indicate that 70% or more of all NK clones are able to recognize certain HLA-B alleles and many also recognize HLA-A alleles. Moreover, a single NK clone has the potential to recognize multiple alleles of HLA-A, HLA-B, and HLA-C antigens. Thus, HLA-C is not unique in conferring protection against NK lysis. (b) No simple patterns of HLA specificity emerged. Examination of a large number of NK clones from a single donor revealed overlapping, yet distinct, patterns of reactivity when a sufficiently broad panel of HLA transfectants was examined. (c) Both autologous and allogeneic HLA antigens were recognized by NK clones. There was neither evidence for deletion of NK clones reactive with self alleles nor any indication for an increased frequency of NK clones recognizing self alleles. (d) With only a few exceptions, protection conferred by transfection of HLA alleles into B-LCL was usually not absolute. Rather a continuum from essentially no protection for certain alleles (HLA-A*0201) to very striking protection for other alleles (HLA-B*5801), with a wide range of intermediate effects, was observed. (e) Whereas most NK clones retained a relatively stable HLA specificity, some NK clones demonstrated variable and heterogeneous activity over time. (f) NK cell recognition and specificity cannot be explained entirely by the presence or absence of HLA class I antigens on the target cell.(ABSTRACT TRUNCATED AT 400 WORDS)