Process auditing using historical data can identify causes for poor performance and reveal opportunities to improve process operation. To date, the data used has been limited to process measurements; however other sources hold complementary information about the process behavior. This paper proposes a new approach to root-cause diagnosis, which also takes advantage of the information in utility, mechanical and electrical data, alarms and diagrams. Its benefit is demonstrated in an industrial case study, by tackling an important challenge in root-cause analysis: large-scale systems. This paper also defines specifications for a semi-automated tool to implement the proposed approach. © 2012 IFAC.