The chemical industry as a whole does not learn from past accidents. Although the individuals concerned and, to some extent, their companies will learn the lessons, the information may not be widely disseminated throughout the industry. Efforts to change this situation have been made over the recent years. Most notably, the reporting of accidents, sometimes anonymously, in various publications. This in turn has led to the creation of computer databases so that appropriate information can be retrieved quickly.This paper identifies the main indexing and retrieval problems with conventional databases. The use of a domain model, in the form of classification hierarchies, is used to improve the process of indexing, identifying and ranking information. Existing databases are on the whole provided as stand alone computer programs and the user must seek out information about accidents by specifying and submitting a query. This paper considers the novel approach of integrating accident databases with computer tools used by chemical plant designers, operators and maintenance engineers, such as CAD systems and digital control systems, so that appropriate accident reports can be automatically retrieved and filtered in response to the context in which the user is working.