This article analyzes the effects of an experimental manipulation of the number of items per screen in a Web survey with forty questions aimed at measuring arousal. The authors consider effects on survey answers, item nonresponse, interview length, and the respondents' evaluation of several aspects of the survey (such as layout). Four different formats are used, with one, four, ten, and forty items and headers on a screen. The authors find no effect of format on the arousal index, but nonresponse increases with the number of items appearing on a single screen. Having multiple items on a screen shortens the duration of the interview but negatively influences the respondent's evaluation of the questionnaire layout. Grouping effects are generally similar for different demographic groups, though there are some differences in magnitude and significance level.