Safe treatment of Plasmodium vivax requires diagnosis of both the infection and status of erythrocytic glucose-6-phosphate dehydrogenase (G6PD) activity because hypnozoitocidal therapy against relapse requires primaquine, which causes a mild to severe acute hemolytic anemia in G6PD deficient patients. Many national malaria control programs recommend primaquine therapy without G6PD screening but with monitoring due to a broad lack of G6PD deficiency screening capacity. The degree of risk in doing so hinges upon the level of residual G6PD activity among the variants present in any given area. We conducted studies on Sumba Island in eastern Indonesia in order to assess the potential threat posed by primaquine therapy without G6PD screening. We sampled 2,033 residents of three separate districts in western Sumba for quantitative G6PD activity and 104 (5.1%) were phenotypically deficient (<4.6U/gHb; median normal 10U/gHb). The villages were in two distinct ecosystems, coastal and inland. A positive correlation occurred between the prevalence of malaria and G6PD deficiency: 5.9% coastal versus inland 0.2% for malaria (P<0.001), and 6.7% and 3.1% for G6PD deficiency (P<0.001) at coastal and inland sites, respectively. The dominant genotypes of G6PD deficiency were Vanua Lava, Viangchan, and Chatham, accounting for 98.5% of the 70 samples genotyped. Subjects expressing the dominant genotypes all had less than 10% of normal enzyme activities and were thus considered severe variants. Blind administration of anti-relapse primaquine therapy at Sumba would likely impose risk of serious harm.