Successful use of e-mail to notify potential respondents of an Internet survey partially depends upon the completeness and accuracy of the e-mail database. In the case of a customer satisfaction survey, when a client's database contains e-mail addresses on only half of its customers, selection bias may occur if only the customers with e-mail addresses are asked to participate. A state agency proposed a customer satisfaction survey using its list of superintendents, business managers and principals within the school districts it serves. Approximately half of the records in the agency's database included an e-mail address for the contact person on record. All records included a postal mailing address. The agency estimated that nearly all of the potential respondents had Internet access in the office even though the agency did not have a complete listing of e-mail addresses. To be sure that all customers had the opportunity to participate, potential respondents were notified about the Internet survey in one of two ways. Respondents with e-mail addresses on file were sent an email notification and respondents without an e-mail address were mailed a letter notifying them about the online questionnaire. A follow-up notification was sent about two-weeks later to most non-responders using the same method as used in the first wave. A subset of randomly selected non-responders who were e-mailed the first notification were sent a second notification via postal mail. Thus, three groups resulted from the two wave notification process: (1) e-mail/e-mail (2) e-mail/paper (3) paper/paper. The response rate effectiveness of these three methods was tracked over time. Differences in response rate and question response by customer type will be examined for each of the methods. Implications for using Internet surveys with paper notifications will be discussed.