The efficacy of a membrane-feeding apparatus as a means of infecting Anopheles dirus mosquitoes with Plasmodium vivax was compared with direct feeding of mosquitoes on gametocyte carriers. Volunteers participating in the study were symptomatic patients reporting to malaria clinics in western Thailand. Direct mosquito feeds were conducted on 285 P. vivax-infected individuals. Four methods of preparing blood for the membrane-feeding apparatus were evaluated. They included 1) replacement of patient plasma with sera from a P. vivax-naive donor (n 276), 2) replacement of patient plasma with plasma from a P. vivax-naive donor (n 83), 3) replacement of patient plasma with that individual’s own plasma (n 80), and 4) whole blood added directly to the feeder (n 221). Criteria used to compare the different methods included 1) number of feeds infecting mosquitoes, 2) percent of mosquitoes with oocysts, and 3) mean number of oocysts per positive mosquito. For most parameters, the direct-feeding method was not significantly different from methods that replaced patient plasma with sera/plasma from a P. vivax-naive donor. However, direct feeding was more effective than use of whole blood or blood that was reconstituted with the patient’s own plasma. These data suggest a possible role of transmission-blocking antibody. The implications towards development of a membrane-feeding assay for the evaluation of candidate transmission-blocking malaria vaccines is discussed.