We conducted a randomized controlled trial to assess the Lifestraw Personal pipe-style water treatment device among a rural population in Ethiopia. A total of 313 households (including 1516 persons) were randomly assigned either to an intervention group in which each householder received a Lifestraw Personal or a control. Households were visited fortnightly over a five-month intervention period and asked to report any episode of diarrhea during the previous week. A random sample of 160 devices was tested each month to assess the presence of thermotolerant coliforms (TTC) and residual iodine in treated water and to measure flow rate under simulated use. Members of the intervention group had 25% fewer weeks with diarrhea than those of the control group (longitudinal prevalence ratio = 0.75; 95% CI 0.60; 0.95). All 718 filtered water samples were free of TTC, were free of detectable iodine disinfectant, and showed a constant flow rate over time. After the five-month intervention period, 34% of participants reported use of device in the preceding week and 13% reported consistent use. While the device was associated with a 25% reduction in longitudinal prevalence of diarrhea, low levels of use suggest that much of this effect is likely to be attributable to reporting bias that is common in open trials with nonobjective outcomes.