Verbal satiation of Chinese bisyllabic words was studied in three experiments to ascertain the phenomenon, to track its time course, and to identify its locus. Experiment 1 asked the participants to judge if an exemplar matched a category in 22 blocks of 40 trials each. Within a block, one category appeared 20 times (repeated trials) while each of the remaining 10 categories appeared only twice (baseline trials). For the first 11 trials, response times (RTs) for the repeated ones were similar to RTs for the baseline ones. For the subsequent trials, repeated RTs were slower (by 9 ms) than baseline RTs, indicating a satiation effect. Its loci could be orthographic, semantic, or both, or on the associative links between form and meaning. In Experiment 2, category names were not shown. Participants judged if two exemplars belonged to the same category. Repeated RTs were faster (by 6 ms) than baseline RTs for the first 12 trials. Then, verbal satiation emerged but was short-lived (between the 13th and the 17th trial) and was of greater magnitude (20 ms) than that observed in Experiment 1. The satiation effect must be semantic, as only meanings were repeated. Experiment 3 asked participants to judge if two category names were identical, mostly an orthographic task. Repeated RTs were similar to baseline ones across all trials, suggesting no orthographic satiation. The results indicate that semantic satiation of Chinese words can be directly semantic (categorical). Its time course conforms to the habituation model described in Rankin, et al. (2009), i.e., sensitization (semantic priming) before habituation (semantic satiation) and habituation followed by dishabituation (recovery).