Web-based surveys increasingly are being used. However, web-based surveys present methodological challenges including lower response rates than other survey methods. This paper reports the results of a large-scale, full factorial experiment to test design features of e-mail and postcard invitations for maximizing web-survey response rates. A stratified, random sample of 13,584 students and faculty of a major U.S. public university received either a postcard or e-mail invitation to participate in a web-based campus sustainability survey. Using a full factorial design, the postcard and e-mail invitations were modified to vary such elements as their length (long/short), estimate of effort (about 10/less than 30 minutes), and subject line (authority/topic). Faculty response (28%/19%, e-mail/postcard) shows evidence of significant effects of design. Postcard invites reduced faculty response probability (p=0.051). Postcards that estimate survey effort at “about 10” minutes increased faculty response (p=0.08). Long e-mail invitations increased faculty response (p=0.018) as did e-mail invitation using the authority subject (“Vice President for Finance and Operations asks you to take a survey”) as opposed to use of the “topic matter” subject (“Take a survey on campus environmental stewardship”) (p=0.022). These results seem to increase the credibility of e-mail for faculty whereas postcards do not need these signals. Student response (14%/13%, e-mail/postcard) did not significantly differ across invitation modes (p=0.36) with invitations estimating effort at “about 10 minutes” increasing student response (p<0.002 for each). For students, time estimate of survey effort mattered. For faculty, e-mail invitations benefited from being longer and from a person of authority apparently signalling some added legitimacy of the invitation. The results suggest that design elements of invitations to maximize response may have different effects on subsets of the target population.