Spaced repetition has been widely implemented and examined in mobile-assisted word learning as an important learning strategy. However, the nature of spaced repetition by commercial word-learning apps and the factors leading to the favoured mobile-assisted spaced repetition have yet to be investigated in authentic contexts. In this study, we coded the spaced repetition patterns and methods of nine apps and interviewed 72 Chinese English learners about their perceptions of spaced repetition for word learning. The results showed three major repetition patterns at three knowledge levels (i.e. the word is unknown, familiar-but-unsure, and known to the learner). The three most common repetition methods were using text-plus-audio for multimedia learning, conducting retrieval practice through flashcards and multiple-choice questions, and integrating game elements such as Goals/Rules and Rewards/Points into learning. Concerning learner preferences, they preferred to have (a) six or seven learning sessions for 'unknown' words, three or four sessions for 'familiar-but-unsure' words, and two or three sessions for 'known' words over ten- to fourteen- day periods, (b) gradually longer intervals between learning sessions, (c) text-plus-audio-plus-image as multimedia, (d) two or three innovative formats of retrieval practice, and (e) integration of Goals/Rules, Rewards/Points, and Time Limits. The results indicate that teachers, researchers, and app designers ought to consider both learning effectiveness and learner perceptions when applying, designing, and developing spaced repetition patterns and methods for mobile-assisted word learning.