Plasmodium vivax is responsible for much of malaria outside Africa. Although P. vivax infections in endemic areas are mostly asymptomatic and associated with very low parasite density, they are deemed important for maintaining malaria transmission. Several studies have shown that a single blood meal from an infected asymptomatic carrier can lead to mosquito infection, but precisely how efficient is not known. To determine the relationship between parasite (gametocyte) density and mosquito infectivity at very low parasitemia, we performed serial dilution of P. vivax-infected blood from malaria patients and used them in membrane feeding assays. Blood samples were collected from P. vivax patients at malaria clinics in Thailand. Thick and thin smears were prepared and parasites and gametocytes densities (number of parasites/gametocytes per microliter blood) were counted under LM. Anopheles dirus females were fed with serially-diluted infected blood through membrane feeders. On day 7 after feeding, mosquitoes were dissected and examined by LM to determine the mosquito infection rate (oocyst prevalence) and the infection intensity (oocyst load). The infection rate and the infection intensity were positively correlated with parasite density. There is large case-to-case variation in parasite transmissibility. The geometric mean parasite density giving 10% mosquito infection rate, was approximately 33 parasites/µl (CI95: 9-120 parasites/µl) or 4 gametocytes/µl (CI95: 1-17 gametocytes/µl). This study demonstrates the ability of P. vivax to infect mosquitoes at very low parasitemia and provides, to date, the most precise relationship between parasite density and mosquito infectivity. The findings provide important information for estimating the contribution of asymptomatic carriers to transmission.