The HLA-DQ alpha allelic variants in the amplified deoxyribonucleic acid (DNA) were determined in a rapid nonradioactive test by hybridization to sequence-specific oligonucleotide probes in both the dot-blot and reverse dot-blot formats. This genetic typing system has been subjected to blind proficiency testing; the performance of this test in the analysis of experimentally mixed samples was also evaluated. As of August 1990, over 250 cases have been tested and more than 2000 individual evidence (bloodstains, semen stains, individual hairs, bone fragments, and tissue sections) and reference samples have been analyzed. The first 198 of these cases are summarized in this paper; in 65 percent of the cases with conclusive results a suspect was included, and in 35 percent, all suspects were excluded. Individual cases as well as some of the general issues relating to forensic science analysis and this genetic typing system are discussed. The high rate of exclusion reported here combined with the ability of PCR to type old evidence samples suggests the relevance of this genetic test for postconviction review; two cases in which the convicted suspect was excluded are discussed. 37 references, 4 tables, and 5 figures (Author abstract)