The survey evidence is interpreted within a Bayseian inference framework. The probability of obtaining the evidence under the hypothesis that the suspect committed the crime is weighed against the probability that the evidence is not associated with the crime. In determining the frequency of occurrence, the definition of population, effect of racial populations of the desired survey, and the rule of mutual independence are considered. The scheme is soundly based on justifiable assumptions and can therefore be defended in court. It avoids unjustifiable assumptions of independence, but does rely on the assumption of independence between different genetic markers within a randomly mating population. A frequency of occurrence can be produced for mixed racial populations in many cases provided the racial composition of the area is known and suitable ethnic blood frequency data are available. 2 tables and 11 references (Author abstract modified)