Multi-label learning and selective ensemble are two hotspot problems in machine learning. Selective ensemble is explored in multi-label learning based on clustering. Two multi-label selective ensemble algorithms， including minimum distance based cluster center selection （MDCCS） and K-means based cluster center selection（KMCCS）， are proposed. The key is to measure the distance between base learners in the proposed algorithms. The learners are represented based on their classification results， then the distance between the learners can be calculated. Besides， two solutions are proposed to solve the problem of empty cluster in the algorithm. Based on the multi-label data sets in Mulan database and five evaluation indexes， the proposed algorithms are analyzed in detail. The experimental results show the effectiveness of the proposed algorithms.