It is valuable to achieve domain adaptation to transfer the learned knowledge from the source labeled CT dataset to the target unlabeled MR dataset for abdominal multi-organ segmentation. Meanwhile, it is highly desirable to avoid high annotation cost of target dataset and protect privacy of source dataset. Therefore, we propose an effective source-free unsupervised domain adaptation method for cross-modality abdominal multi-organ segmentation without accessing the source dataset. The process of the proposed framework includes two stages. At the first stage, the feature map statistics-guided model adaptation combining with entropy minimization is developed to help the top segmentation network to achieve reliable segmentations on the target images. The pseudo-labels outputted from the top segmentation network is used to guide the style compensation network to generate source-like images. The pseudo-labels outputted from the middle segmentation network is used to supervise the learning of the desired model (the bottom segmentation network). At the second stage, the circular learning and the pixel-adaptive mask refinement are used to further improve the performance of the desired model. With this approach, we achieve satisfactory performance on the abdominal multi-organ segmentation, outperforming recent state-of-the-art domain adaptation methods. The proposed approach can also be easily extended to the situation when there exists target annotation data. With only one labeled MR volume, the performance is improved to close to supervised learning. Furthermore, the proposed approach is proved to be effective in source-free unsupervised domain adaptation in reverse direction.