This paper investigates trust in the scientists, government authorities and wider risk management team during the ongoing volcanic crisis in Montserrat, WI. Identifying the most trusted communicator and how trust in information can be enhanced are considered important for improving the efficacy of volcanic risk communication. Qualitative interviews, participant observations and a quantitative survey were utilised to investigate the views and attitudes of the public, authorities and scientists. Trust was found to be dynamic, influenced by political factors made more complex by the colonial nature of Montserrat’s governance and the changing level of volcanic activity. The scientists were viewed by the authorities as a highly trusted expert source of volcanic information. Mistrust among some of the local authorities towards the scientists and British Governor was founded in the uncertainty of the volcanic situation and influenced by differences in levels of acceptable risk and suspicions about integrity (e.g. as a consequence of employment by the British Government). The public viewed friends and relatives as the most trusted source for volcanic information. High trust in this source allowed competing messages to reinforce beliefs of lower risk than were officially being described. The scientists were the second most trusted group by the public and considered significantly more competent, reliable, caring, fair and open than the authorities. The world press was the least trusted, preceded closely by the British Governor’s Office and Montserratian Government officials. These results tally well with other empirical findings suggesting that government ministers and departments are typically distrusted as sources of risk-related information. These findings have implications for risk communication on Montserrat and other volcanic crises. The importance and potential effectiveness of scientists as communicators, because of, and despite, the existence of political, cultural and institutional barriers, is exemplified by this study.