Although a deep body of literature exists on the design of web surveys of the general population, little or no studies have examined the design of such surveys for individuals with disabilities. Well-intentioned researchers often design web surveys that fall short of compliance with Section 508 of the U.S. Rehabilitation Act and rarely go beyond compliance to consider maximizing the accessibility of their data collection tools. The current study presents the design challenges faced and strategies developed in creating and testing a web survey target ingthose eligible for the Talking Book and Braille Program through the Library of Congress National Library Service for the Blind and Physically Handicapped (NLS). Many of the standard practices in designing web surveys, such as minimizing scrolling on web pages, are not effective in designing surveys for respondents with visual disabilities. Furthermore, taking a web survey with a screen reader often takes longer than taking it without one, increasing respondent burden for those using screen readers. Testing web surveys for accessibility with multiple screen readers can also be challenging for researchers who are not familiar with the technologies and do not have access to the numerous different technologies available. Commercial and free screen reader software is available, but these can often be difficult for non-users to master which can create challenges in testing for ease of use and response burden. Also important is testing the survey with those who will be using other methods of web accessibility, such as magnification software. Data sources include participant reactions captured during a pilot test of the web survey and anecdotally reported throughout the survey period, as well as a comparative analysis of average survey length among respondents using the web survey versus average survey length among those completing the same survey over the phone.