Abstract This paper presents an evaluation method for a multi-chiller system’s operation performance. It develops the near-optimal performance map as the evaluation benchmark covering all different cooling conditions of the multi-chiller system. For that map, a new data-based chiller power model using a dedicated set of coefficients for each of different cooling conditions is developed to supplement load-power data into the history operation data of the multi-chiller system. According to the evaluation benchmark, an evaluation index I s ˜ , is defined to show the potential improvement in the operation performance of the multi-chiller system. The near-optimal performance map is also used to analyze the distribution of the multi-chiller system’s operation conditions. To validate the proposed evaluation method, three operation strategies for an existed multi-chiller system are evaluated. The data-based chiller power model is proved accurate enough in the use of the evaluation. According to the evaluation results, the three strategies are compared and advised. The validation demonstrates that the proposed method is capable to evaluate the operation performance of the multi-chiller system.