Experiments were conducted in two cities comparing web and mail surveys seeking resident attitudes about city services, quality of community life and local policies. Web participants were recruited the same way as mail participants, through a probability sample of addresses, but web recruits were mailed the survey URL instead of the questionnaire. Response rates, respondent demographic characteristics, use of "Don't Know" and mean responses as well as response variability were compared. Although in each city, response rates were much lower for the web participants, and web respondents had different demographic characteristics and tendencies to be opinionated compared to mail respondents, the variability and means of results were quite similar for the two survey groups. Differences between web responses and mail responses also were similar in both cities. Despite coverage limitations, web administration may have limited utility for citizen surveys where political and technical barriers can be overcome.