This study investigated the effects of temporal distribution of practice on the learning and retention of Mandarin tonal word production. Eighty native English-speaking adults with no prior knowledge of a tonal language participated in this study. They were randomly assigned to four experimental conditions varying in intersession interval (ISI) and retention interval (RI): Condition A with a 1-day ISI and a 1-week RI, Condition B with a 1-day ISI and a 4-week RI, Condition C with a 1-week ISI and a 1-week RI, and Condition D with a 1-week ISI and a 4-week RI. Participants underwent 3 training sessions followed by a final retention test. Results showed that the effects of ISI and RI differed depending on the type of knowledge to be retained. For the retention of declarative knowledge, RI had a robust effect: the longer the RI, the worse the retention. Compared to shorter spaced practice (1-day ISI), longer spaced practice (1-week ISI) seemed to improve long-term retention (4-week RI) of declarative knowledge; however, this ISI effect was much weaker. For procedural knowledge retention, ISI played an important role, but not RI, and it was the shorter spaced practice schedule that had an advantage over the longer spaced practice schedule.