The lack of Design for Maintainability (DfM) leads to avoidable building component defects and unproductive building operations. This study reports on the development of a building design assessment system which predicts maintainability impacts of design alternatives. A mixed method approach was used in this study where 1372 DfM benchmarks from 404 national and international building standards were analysed qualitatively to identify benchmarks relating to 123 critical building defects. Five Bayesian Belief Networks (BBN) were developed using five-year defect data and expert judgements from 95 case buildings. Monetary quantification of results was carried out, indicating the complex relationships between DfM benchmarks and building defects to great effect. The Bayesian networks were operationally validated using scenario tests and sensitivity analysis. The predictive power of the developed system is illustrated using an application to a hypothetical case study. This novel consequence-based DfM assessment system is a decision making tool aiding in designing more maintainable building systems which are more productive in terms of materials, labour and cost.