The current challenge of maintenance and asset management in continuous process industries is to anticipate failures to avoid downtime and consequent production losses. Consequently, new knowledge, which may be obtained from the exploitation of data stored in computerized systems, is required to support this function. The extraction of knowledge from a database for strategic business purposes, however, is not obtained directly, it is necessary to use appropriate and effective methodologies and tools. In this sense, a case study was developed in a pulp industry in the northeastern region of Brazil with the objective of applying the knowledge discovery in databases methodology to identify opportunities to improve maintenance management. The obtained results present important discoveries such as the precise identification of the main areas, processes, maintenance disciplines and, above all, the equipment that most impact in terms of frequency of anomalies and production losses over the years investigated. Through classifying algorithms, it is also possible to obtain missing data, so important for the improvement of information and analysis of the maintenance function.
