OBJECTIVES: Insecticide-treated bednets (ITNs) are effective in preventing nocturnally transmitted vector-borne diseases, but their effect on diurnally active dengue vectors has never been studied. We investigated the efficacy of ITNs in reducing Aedes aegypti populations and dengue transmission. METHODS: A cluster-randomized trial was carried out in Leogane, Haiti between July 2003 and July 2004. The study area (1017 houses) was divided into 18 sectors (clusters): nine received ITNs (Olyset(R) long-lasting insecticidal bednets) and nine were untreated controls. Entomological surveys [measuring Breteau (BI), house (HI), container (CI) and pupae per person (PPI) indices and oviposition activity] were undertaken at baseline and at 1 and 5 months post-intervention. All houses were georeferenced to enable spatial analysis. Control sectors received ITNs at 6 months, and a final entomological and attitudinal survey was undertaken at 12 months after baseline. Anti-dengue IgM seropositivity rates were measured at baseline and after 12 months. Efficacy of ITNs was assessed by WHO cone bioassays. RESULTS: At 1-month post-intervention, entomological indices fell in all sectors, with HI and BI in the bednet sectors reduced by 6.7 (95% CI -10.6, -2.7; P < 0.01) and 8.4 (95% CI -14.1, -2.6; P < 0.01) respectively. Moreover at 1 month, ovitraps in control sectors were significantly more likely to be positive than in bednet sectors (P < 0.01). By 5 months, all indices remained low and HI, CI and BI were also significantly lower than that of baseline in the control arm. Curiously, at 5 months, HI, CI and BI were lower in the control arm than that in the bednet arm. A final survey, 12 months after the initial baseline study (5 months after bednets had been given to all households) indicated that all indices were significantly lower than that at baseline (P < 0.001). Control houses located within 50 m of a bednet house had significantly lower CI (Z = -2.67, P = 0.008) and PPI (Z = -2.19, P = 0.028) at 1 month, an effect that extended to 100 m by 5 months (Z = -2.03, P = 0.042 and Z = -2.37, P = 0.018 respectively), suggesting a spill-over effect of the bednets. An IgM serosurvey showed a 15.3% decrease (95% CI 5.0-25.5%, P < 0.01) in the number of IgM-positive individuals from baseline to12 months later. CONCLUSIONS: Insecticide-treated bednets had an immediate effect on dengue vector populations after their introduction, and over the next 5-12 months, the presence of ITNs may have continued to affect vector populations and dengue transmission.