BACKGROUND: Phlebotomines are a group of insects which include vectors of the Leishmania parasites that cause visceral leishmaniasis (VL) and cutaneous leishmaniasis (CL), diseases primarily affecting populations of low socioeconomic status. VL in Brazil is caused by Leishmania infantum, with transmission mainly attributed to Lutzomyia longipalpis, a species complex of sand fly, and is concentrated mainly in the northeastern part of the country. CL is distributed worldwide and occurs in five regions of Brazil, at a higher incidence in the north and northeast regions, with etiological agents, vectors, reservoirs and epidemiological patterns that differ from VL. The aim of this study was to determine the composition, distribution and ecological relationships of phlebotomine species in an Atlantic Forest conservation unit and nearby residential area in northeastern Brazil. METHODS: Centers for Disease Control and Shannon traps were used for collections, the former at six points inside the forest and in the peridomestic environment of surrounding residences, three times per month for 36 months, and the latter in a forest area, once a month for 3 months. The phlebotomines identified were compared with climate data using simple linear correlation, Pearson's correlation coefficient and cross-correlation. The estimate of ecological parameters was calculated according to the Shannon-Wiener diversity index, standardized index of species abundance and the dominance index. RESULTS: A total of 75,499 phlebotomines belonging to 11 species were captured in the CDC traps, the most abundant being Evandromyia walkeri, Psychodopygus wellcomei and Lu. longipalpis. Evandromyia walkeri abundance was most influenced by temperature at collection time and during the months preceding collection and rainfall during the months preceding collection. Psychodopygus wellcomei abundance was most affected by rainfall and relative humidity during the collection month and the month immediately preceding collection time. Lutzomyia longipalpis abundance showed a correlation with temperature and the rainfall during the months preceding collection time. The Shannon trap contained a total of 3914 phlebotomines from these different species. Psychodopygus wellcomei, accounting for 91.93% of the total, was anthropophilic and active mainly at night. CONCLUSIONS: Most of the species collected in the traps were seasonal and exhibited changes in their composition and population dynamics associated with local adaptions. The presence of vectors Ps. wellcomei and Lu. longipalpis underscore the epidemiological importance of these phlebotomines in the conservation unit and surrounding anthropized areas. Neighboring residential areas should be permanently monitored to prevent VL or CL transmission and outbreaks.