BackgroundInsecticide spraying efficiently controls house infestation by triatomine bugs, the vectors of Trypanosoma cruzi. The strategy, however, is ineffective against sylvatic triatomines, which can transmit Chagas disease by invading (without colonizing) man-made structures. Despite growing awareness of the relevance of these transmission dynamics, the drivers of house invasion by sylvatic triatomines remain poorly understood.Methods/FindingsAbout 12,000 sylvatic triatomines were caught during routine surveillance in houses of Tocantins state, Brazil, in 2005–2013. Using negative binomial regression, information-theoretic model evaluation/averaging, and external model validation, we investigated the effects of regional (Amazon/Cerrado), landscape (preservation/disturbance), and climate covariates (temperature, rainfall) on the municipality-aggregated numbers of house-invading Rhodnius pictipes, R. robustus, R. neglectus, and Panstrongylus geniculatus. House invasion by R. pictipes and R. robustus was overall more frequent in the Amazon biome, tended to increase in municipalities with more well-preserved land, and decreased in rainier municipalities. Across species, invasion decreased with higher landscape-disturbance levels and in hotter-day municipalities. Invasion by R. neglectus and P. geniculatus increased somewhat with more land at intermediate disturbance and peaked in average-rainfall municipalities. Temperature effects were more pronounced on P. geniculatus than on Rhodnius spp.ConclusionsWe report widespread, frequent house invasion by sylvatic triatomines in the Amazon–Cerrado transition. Our analyses indicate that readily available environmental metrics may help predict the risk of contact between sylvatic triatomines and humans at coarse geographic scales, and hint at specific hypotheses about climate and deforestation effects on those vectors–with some taxon-specific responses and some seemingly general trends. Thus, our focal species appear to be quite sensitive to higher temperatures, and might be less common in more heavily-disturbed than in better-preserved environments. This study illustrates, in sum, how entomological routine-surveillance data can be efficiently used for Chagas disease risk prediction and stratification when house-colonizing vectors are absent.