West Nile virus (WNV) can cause large outbreaks of febrile illness and severe neurologic disease. This study estimates the seroprevalence of WNV infection and assesses risk perception and practices regarding potential exposures to mosquitoes of persons in an area with intense epizootics in 1999 and 2000. A serosurvey of persons aged > or = 12 years was conducted in southwestern Connecticut during October 10-15, 2000, using household-based stratified cluster sampling. Participants completed a questionnaire regarding concern for and personal measures taken with respect to WNV and provided a blood sample for WNV testing. Seven hundred thirty persons from 645 households participated. No person tested positive for WNV (95% CI: 0-0.5%). Overall, 44% of persons used mosquito repellent, 56% practiced > or = two personal precautions to avoid mosquitoes, and 61% of households did > or = two mosquito-source reduction activities. In multivariate analyses, using mosquito repellent was associated with age < 50 years, using English as the primary language in the home, being worried about WNV, being a little worried about pesticides, and finding mosquitoes frequently in the home (P<0.05). Females (OR = 2.0; CI = 1.2-2.9) and persons very worried about WNV (OR = 3.8; CI = 2.2-6.5) were more likely to practice > or = two personal precautions. Taking > or = two mosquito source reductions was associated with persons with English as the primary language (OR = 2.0; CI = 1.1-3.5) and finding a dead bird on the property (OR = 1.8; CI = 1.1-2.8). An intense epizootic can occur in an area without having a high risk for infection to humans. A better understanding of why certain people do not take personal protective measures, especially among those aged > or = 50 years and those whose primary language is not English, might be needed if educational campaigns are to prevent future WNV outbreaks.