Reliability of production processes is a key issue to ensuring a stable system operation, increasing a product quality, and reducing production losses. In this paper we proposed a tool for the analysis of faults in a process and the definition of the most effective way for their elimination. In the centre of the offered framework is FMEA - a reliability analysis type, the most widely used in enterprises. In the paper it is proposed to extend the FMEA by introducing a classification of faults and an estimation of expert opinions for the FMEA parameters. By using the Pareto analysis, it is possible to extract from the FMEA the most critical process failures. To analyse these faults through the Bayesian Belief Network is the most effective way to address them. The Bayesian network structure duplicates the faults classifier structure, therefore this method fits well for this analysis. BBN enables to calculate the probabilities for each fault group based on the error of the manufacturing processes probability. For a more complete analysis of a process we used a structural and dynamic analysis for revealing the bottlenecks of the process, as well as the Fault Tree Analysis and Reliability Block Diagram, which may be built on the base of a structural model of a process and gives the reliability of the system on the whole. On the base of the FMEA data it is possible to calculate an optimal plan of the equipment maintenance for a current process. The framework for the analysis of the production process enables companies of machinery manufacturing enterprises to analyse processes as a whole as well as its parts for efficient forecast of the production process improvement.DOI: http://dx.doi.org/10.5755/j01.mech.18.6.3168