Using 18 different blood group systems, 2,124 cases of doubtful paternity, unevenly distributed among four different population groups--South African Negroes (Xhosas) 645 (30%); South African Caucasians 264 (13%); Cape Coloureds (non-Malay) 1,156 (54%), and Cape Malays 59 (3%)--were analysed. Tests for red cell antigen, plasma protein, red cell enzyme and HLA polymorphisms were done on all subjects. The overall exclusion rate was 38,23%, i.e. 812 of the 2,124 men were not biological fathers of the children ascribed to them. The capability of the individual systems to exclude a man from specific paternity was also evaluated. The polymorphic HLA system was the most useful and alone--depending on the particular population group--excluded between 93.5% and 97.8% of falsely accused men. The proportion of men excluded also varied between different population groups using the different systems, e.g. the Rhesus system alone excluded 43.9% of South African Caucasian non-fathers and only 14% of the Xhosa non-fathers. This is a reflection of the gene frequencies within a particular population. The analysis showed that the genetic systems tested in this laboratory were extremely efficient in providing evidence in proof of non-paternity.