Although energy piles are widely used worldwide as green foundations, currently no approach is available to quantify the design uncertainties in their long-term performance. Based on the cyclic load-transfer curves, this paper presents a probabilistic model for life-cycle analysis of energy piles under seasonal thermal loading. For computational efficiency, the subdomain sampling method is adopted for estimating the failure probability due to excessive settlement, and is compared with Monte Carlo simulation. This probabilistic analysis approach for energy piles is illustrated through a case study. The life-cycle failure probability of energy piles for various mechanical loading and thermal loading scenarios is investigated. (C) 2019 Elsevier Ltd. All rights reserved.