Replacement of air-cooled chillers with water-cooled chillers for air-conditioning plants in existing buildings can yield a significant amount of energy savings, especially in a sub-tropical climate. However, due to variations in weather conditions and building operation patterns, the amount of actual energy saving is often uncertain in these retrofits. These uncertainties impose a risk of saving shortfalls when Energy Service Companies (ESCOs) guarantee building owners for a certain amount of energy savings in Energy Performance Contracting (EPC) projects. This study presents a probabilistic approach to estimating a range of possible energy savings with the associated confidence levels for chiller replacement in existing buildings, taking into account the annual variations in the influential parameters affecting energy savings. The influential factors include building cooling loads, system control and operation systems, as well as chiller plant characteristics. The proposed approach involves: the use of correlation analysis for identifying influential parameters; EnergyPlus for simulating energy use of chiller plant; and a Monte Carlo approach for simulating the probability of post-retrofit energy savings. A commercial building where the air-cooled chillers were replaced with water-cooled chillers is used to illustrate the proposed approach. Results show that the variations in annual energy savings for chiller replacement projects can be estimated with a defined degree of certainty. In the case study project, the possible annual energy savings during the post-retrofit period range from 1,149,000 kWh (37.6% of baseline consumption) to 1,504,000 kWh (49.2% of ditto) at 90% statistical significance. The risk mitigation measures for this type of energy retrofit are discussed as well. (C) 2016 Elsevier Ltd. All rights reserved.