We used a simple HLA-B35 PCR/SSO-oligotyping procedure, combined with exon 3 direct sequencing for the analysis of B35 subtype frequencies in our population, and for the evaluation of the degree of B35-subtype compatibility in serologically matched unrelated bone marrow and kidney transplant pairs. B*3501 was the most frequent allele (0.6), followed by B*3503 (0.19), B*3502 (0.13), B*3508 (0.07), and B*3505 (< 0.01). HLA-B35-subtype matching of donors and recipients was strongly dependent on the stringency of ABDRB1 matching. Among 10 kidney donor/recipient pairs, only 30% were B35-subtype-matched. Due to the lack of ABDRB1 haplotype matching, this low degree of matching was not better than what would be expected on the basis of the subtype frequency distribution in the population. In contrast, HLA-B35 subtype compatibility was higher in unrelated bone marrow donor/recipient pairs that were serologically ABDR-matched: 30 of the 62 (48.4%) B35-positive combinations tested were B35-subtype-compatible. When all patient/donor plus donor/donor combinations (n = 160) were taken into account, 46% of the ABDR-matched pairs were B35-subtype-compatible. When only pairs that were DRB1/DRB3/DRB5-subtype-matched by oligotyping (n = 62) were considered, 71% were B35-subtype-compatible. The fact that a significant percent of patient/donor pairs matched by conventional HLA-typing are found incompatible, as shown here for B35, explains the difficulties in assessing the beneficial effect of HLA matching in transplantation.