This paper examines a potential issue in measuring service quality using the SERVQUAL instrument. Specifically, it is argued that SERVQUAL’s dimensions are largely the product of presentation order effects. To validate this argument, this paper presents the results of a field study in which randomised and non-randomised versions of SERVQUAL were administered in multiple organisations. The resulting samples were then used to generate factor structures which proved to be non-congruent. These results raise the question of whether SERVQUAL’s dimensions truly represent underlying constructs. If they do not, then should studies continue to report them?