The populations fell into two groups - highland and coastal. Comparisons were mainly between individual populations, classified linguistically, within each group. Calculated gene frequencies derived from the results on 12 systems (5 red cell antigen, 2 serum protein and 5 red cell enzyme), which showed useful variation, were used to estimate pair-wise genetic distances. Cluster analysis showed good correlation between linguistic and genetic information. The assumption that drift rather than selection had been the main cause of the observed variations between the populations studied seemed justified by comparisons of time scales produced by lexico-statistical methods with those derived from genetic distances. One particular measure of variation (f<latex>$_\theta$</latex>) may be useful in providing some approximate and tentative estimate of the times of separation of related New Guinean population groups.