Abstract-The computer security community has advocated widespread adoption of secure communication tools to protect personal privacy. Several popular communication tools have adopted end-to-end encryption (e.g., WhatsApp, iMessage), or promoted security features as selling points (e.g., Telegram, Signal). However, previous studies have shown that users may not understand the security features of the tools they are using, and may not be using them correctly. In this paper, we present a study of Telegram using two complementary methods: (1) a labbased user study (11 novices and 11 Telegram users), and (2) a hybrid analytical approach combining cognitive walk-through and heuristic evaluation to analyse Telegram's user interface. Participants who use Telegram feel secure because they feel they are using a secure tool, but in reality Telegram offers limited security benefits to most of its users. Most participants develop a habit of using the less secure default chat mode at all times. We also uncover several user interface design issues that impact security, including technical jargon, inconsistent use of terminology, and making some security features clear and others not. For instance, use of the end-to-end-encrypted Secret Chat mode requires both the sender and recipient be online at the same time, and Secret Chat does not support group conversations.