This thesis looks at the issue of interdomain user authentication, i.e. user authentication in systems that extend over more than one administrative domain. It is divided into three parts. After a brief overview of related literature, the first part provides a taxonomy of current approaches to the problem. The taxonomy is first used to identify the relative strengths and weaknesses of each approach, and then employed as the basis for putting into context four concrete and novel schemes that are subsequently proposed in this part of the thesis. Three of these schemes build on existing technology; the first on 2nd and 3rd-generation cellular (mobile) telephony, the second on credit/debit smartcards, and the third on Trusted Computing. The fourth scheme is, in certain ways, different from the others. Most notably, unlike the other three schemes, it does not require the user to possess tamper-resistant hardware, and it is suitable for use from an untrusted access device. An implementation of the latter scheme (which works as a web proxy) is also described in this part of the thesis. As the need to preserve one’s privacy continues to gain importance in the digital world, it is important to enhance user authentication schemes with properties that enable users to remain anonymous (yet authenticated). In the second part of the thesis, anonymous credential systems are identified as a tool that can be used to achieve this goal. A formal model that captures relevant security and privacy notions for such systems is proposed. From this model, it is evident that there exist certain inherent limits to the privacy that such systems can offer. These are examined in more detail, and a scheme is proposed that mitigates the exposure to certain attacks that exploit these limits in order to compromise user privacy. The second part of the thesis also shows how to use an anonymous credential system in order to facilitate what we call ‘privacy-aware single sign-on’ in an open environment. The scheme enables the user to authenticate himself to service providers under separate identifier, where these identifiers cannot be linked to each other, even if all service providers collude. It is demonstrated that the anonymity enhancement scheme proposed earlier is particularly suited in this special application of anonymous credential systems. Finally, the third part of the thesis concludes with some open research questions.