In recent years, the number of emails received by an individual has been increasing, and the time required for email triage (i.e., the process of going through unhandled emails and deciding what to do with them) has therefore been increasing. Golbeck et al. proposed TrustMail, which is a prototype email client that prioritizes emails in user’s mailbox utilizing a trust network (i.e., a social network representing trust relationships among users). In this paper, we extend the notion of TrustMail to allow robust message-based prioritization using inter-recipient trusts, which are (inferred) trust scores from the recipient to other recipients. We propose a method called EMIRT (Estimating Message Importance from inter-Recipient Trust) for enabling robust message prioritization. We also evaluate the effectiveness of our proposed EMIRT for estimating importance of emails through experiments utilizing a large email corpus called Enron Email Dataset. Consequently, we show that EMIRT realizes robust email prioritization.