Many users are now significantly dependent upon computer application. Whilst many aspects are now used very successfully, an area in which usability difficulties continue to be encountered is in relation to security. Thus can become particularly acute in situations where users are required to interact and make decisions, and a key context here is typically when they need to respond to security warnings. The current implementation of security warnings can often be considered as an attempt to offer a one size fits all solution. However, it can be argued that many implementations are still lacking the ability to provide meaningful and effective warnings. As such, this research focuses upon achieving a better understanding of the elements that aid end-users in comprehending the warnings, the difficulties with the current approaches, and the resulting requirements in order to improve the design and implementation of such security dialogues. In the early stage of research, a survey was undertaken to investigate perceptions of security dialogues in practice, with a specific focus upon security warnings issued within web browsers. This provided empirical evidence of end-users’ experiences, and revealed notable difficulties in terms of their understanding and interpretation of the security interactions. Building upon this, the follow-up research investigated understanding of application level security warnings in wider contexts, looking firstly at users’ interpretation of what constitutes a security warning and then at their level of comprehension when related warnings occurred. These results confirmed the need to improve the dialogues so that the end-users are able to act appropriately, and consequently promoted the design and prototype implementation of a novel architecture to improve security warnings, which has been titled Automated Security Interface Adaptation (ASIA). The ASIA approach aims to improve security warnings by tailoring the interaction more closely to individual user needs. By automatically adapting the presentation to match each user’s understanding and preferences, security warnings can be modified in ways that enable users to better comprehend them, and thus make more informed security decisions and choices. A comparison of the ASIA-adapted interfaces compared to standard versions of warnings revealed that the modified versions were better understood. As such, the ASIA approach has significant potential to assist (and thereby protect) the end-user community in their future interactions with security.