Extreme learning machine (ELM) has been well recognized as a new learning scheme for single-hidden layer feedforward networks (SLFNs) with extremely fast learning speed and high generalization performance. However, the stability and accuracy of ELM can be further enhanced. In this paper, a new hybrid machine learning method called robust AdaBoost.RT based ensemble ELM (RAE-ELM) for regression problems is proposed, which combined ELM with the novel self-adaptive AdaBoost.RT algorithm to achieve better approximation accuracy than using only single ELM network. To enable dynamically self-adjusting the threshold value during AdaBoost.RT computing without any empirical suggestion, the statistical parameters of applying ELM networks on the input dataset are adopted as indicators. The experiment results that applying the proposed algorithm on wide types of benchmark databases verify that RAE-ELM not only outperforms the conventional ELM but also achieves a higher stability and accuracy level than original and modified AdaBoost.RT-based ELM for regression problems.