Development of long-lasting insecticidal nets (LLINs) may eliminate the need for insecticide retreatment of ITNs. While two LLINs (Olyset, Sumitomo Chemical Co., Japan; and PermaNet 1.0, Vestergaard-Frandsen, Denmark) have received recommendations from the World Health Organization Pesticide Evaluation Scheme, field-testing under normal use has been limited. We used a survival analysis approach to compare time to net failure of conventional polyester bednets treated only with deltamethrin to two LLINs and two candidate LLINs (Olyset; PermaNet; Insector, Athanor, France; and Dawa, Siamdutch Mosquito Netting Co., Thailand). Additionally, we evaluated nets treated with a process designed to increase the wash-durability of permethrin-treated nets through the addition of cyclodextrin (a starch) in the treatment process. Houses in western Kenya were randomly assigned to one of the six net types and nets were distributed to cover all sleeping spaces. Households were visited monthly to assess reported side effects in inhabitants and washing frequency. Nets were evaluated for insecticidal activity by periodic WHO cone bioassays with mortality assessed at 24 h. Nets with bioassay mortality <70% were assayed monthly until failure, defined as the first of two consecutive bioassay mortality rates <50%. Time to failure was analyzed using an extended Cox Proportional Hazards model controlling for the cumulative number of washes. We distributed 314 nets to 177 households in June-July 2002; 22 nets (7.0%) were lost to follow-up and 196 (62.4%) failed during the first 2 years of the evaluation. Controlling for cumulative number of washes, PermaNet 1.0 [Hazard Ratio (HR) 0.14, 95% Confidence Interval (CI) 0.06-0.31] had a significantly lower risk of failure than conventional nets while Insector had a significantly higher risk of failure (HR 2.57, 95% CI 1.06-4.15). The risks of failure of the remaining nets (Olyset: HR 1.29, 95% CI 0.79-2.10; Dawa: HR 0.58, 95% CI 0.32-1.18; cyclodextrin: HR 0.65, 95% CI 0.40-1.1) were not significantly different from that of a conventional net. PermaNet 1.0 performed significantly better than conventional nets and should be recommended to malaria control programs.