Basic telephone service has long been defined as universal service, access to which is guaranteed across the entire nation by regulatory systems such as a universal service fund that collects contributions indirectly from general telecom users to compensate for the costs of providing the service to disadvantaged regions. As broadband spreads nationwide and telecom networks become broadband and IP-based, both regulatory bodies and academics have suggested that the definition of universal service may need to be changed to guarantee connectivity rather than a specific service.To assess whether the existing system is still appropriate for newly-defined universal service, not only must its social benefits exceed the costs in terms of cost-benefit analysis, but also the method of charging telecom users for universal service should be consistent with the attributes and preferences because these have an impact on the benefits.This paper investigates the relationships between the benefits of the universal service system and people's attributes and preferences in Japan, assuming that broadband is designated as a universal service. Benefits were estimated as users' willingness-to-pay (WTP) from web survey data by a contingent valuation method (CVM) using a logit model based on the random utility theory. Preference for the universal service system was measured as the reasons respondents said they were willing to pay for the assumed universal service, i.e. the economic rationales (such as network externality, redistribution, regional policy and merit goods) as well as self-interest.The estimation showed that the median WTP for broadband as universal service is approximately 100 JPY. In terms of relationships with the benefits, the analysis suggested that the self-interest of inhabitants and visitors in disadvantaged regions significantly affects their WTP. Also, redistribution among regions, which more than half of the respondents stated was the reason they supported the universal service system, was found to significantly affect WTP. This suggests that redistribution justifies a new universal service system in Japan. Other rationales or user attributes including income, communications expense, sex and age did not appear to have a significant influence. From these results, implications are drawn on how future universal service regulation should be handled, despite some limitations associated with the survey method. Finally, challenges for future studies on universal access are mentioned.JEL Classification: L96, L51, D61