To accommodate applications with highly variable degrees and categories of Quality of Service (QoS) requirements, a major effort has been invested to overcome the traditional best-effort service quality of the Internet [1],[2]. Nevertheless, there has been a shift towards weak QoSassurance paradigms based on the differentiation of different QoS levels [3]. What is currently missing is a technique for specifying user preferences for QoS and network charging mechanisms. If the Internet is to remain economically viable, demand for Internet services must be maintained. Specifying users' QoS requirements therefore has potentially crucial implications for providing a degree of service that will encourage users to pay for the quality that they receive. This paper investigates the role of conceptual design in establishing users' mental representations of Internet services and their charges. We report empirical work that shows how these models can be used to predict the acceptability ...