In this paper, an online self-tuning generalized minimum variance (GMV) controller is proposed for a 100 KW waste heat recovery system with organic Rankine cycle (ORC). The ORC process model is formulated by the controlled autoregressive moving average (CARMA) model whose parameters are identified using the recursive least squares (RLS) algorithm with forgetting factor. The generalized minimum variance algorithm is applied to regulate ORC based waste heat recovery system. The contributions of this work are twofold: (1) the proposed control strategy is formulated under the data-driven framework, which does not need the precise mathematic model; (2) this proposed method is applied to handle tracking set-point variations and process disturbances by improved minimum objective GMV function. The performance of GMV controller is compared with the PID controller. The simulation results show that the proposed strategy can achieve satisfactory set-point tracking and disturbance rejection performance. (C) 2014 Elsevier Ltd. All rights reserved.