Visual analogue scales (VASs) are graphic rating scales especially suited for selfadministered Web surveys. Theoretical reasoning and findings from paper-based research advocate the use of VASs that promise data on the interval level and a reliable measurement. Nevertheless, VASs lead a shadowy existence, mainly being used in medical research but seldom for surveying the general population. Even though VASs lend themselves for computerized assessment, only few studies have been conducted on the Web (e.g. Couper, Tourangeau, Conrad, & Singer 2006; Reips & Funke 2008; Turpin et al. 2003; van Schaik & Ling 2007). This may be because those in charge of designing surveys are unsure whether respondents can use this kind of rating scale in a meaningful way. To resolve the uncertainty, we report a systematic simulation and two empirical studies on the error following from the use of VASs in comparison to error following from the use of categorical scales. More precisely, we examine formatting error, the difference between true and reported value caused by the available rating scale.