Despite the low-carbon environmental policies, coal is expected to remain a main source of energy in the coming decades. Therefore, efficient and environmentally friendly power systems are required. A design process based on the deterministic models and application of the safety factors leads to the equipment oversizing, hence fall in the efficiency and increase in the capital and operating costs. In this work, applicability of a non-intrusive stochastic methodology to determine the probability of the power plant equipment failure was investigated. This alternative approach to the power plant performance assessment employs approximation methods for the deterministic prediction of the key performance indicators, which are used to estimate reliability indices based on the uncertainty of the input to a process model of the coal-fired power plant. This study revealed that high reliability indices obtained in the analysis would lead to reduced application of conservative safety factors on the plant equipment, which should result in lower capital and operating cost, through a more reliable assessment of its performance state over its service time, and lead to the optimisation of its inspection and maintenance interventions. (C) 2014 Elsevier Ltd. All rights reserved.