Children and adults are able to learn a motor sequence quickly, usually over a course of one learning session consisting of 4-8 learning blocks. This initial acquisition is referred to as fast learning. However, little is known about the learning processes underlying the fast acquisition of motor sequences. Therefore, the overarching objective of this dissertation was to examine the underlying processes that drive rapid motor sequence learning in children and adults. In a series of studies, children and adults performed a modified serial reaction time (SRT) task, a primary window into understanding implicit motor sequence learning. Study I demonstrated that fast learning of implicit motor sequences in six- and 10-year-old children was comparable to adults, while the performance (i.e., reaction time, RT) during learning was reflected by two age-related processes. Learning in six-year-old children dominantly relied on an offline process where RT improved after a short rest, while offline enhancement as well as online progressive improvement in RT reflected sequence learning in 10-year-old children and adults. In studies II, III, and IV, we demonstrated that the online and offline processes were neither by-products of task pacing constraints nor illusory effects of fatigue or reactive inhibition. Instead, these two age-related processes were more likely to be functional mechanisms underlying implicit motor sequence learning, which could be modulated by the involvement of procedural and declarative memory. In addition, study III characterized the developmental landscape of 5- to 14-year-old children and found that the developmental changes of online and offline learning were primarily present in early childhood. As fast learning is known to enable generalization (or transfer) of sequences learning, we expected, given the findings in studies I through IV, age-related differences in the generalization of implicit motor sequence learning. The results in study V, interestingly, demonstrated that the generalization of implicit motor sequence learning was better in children than in adults. However, in study VI, when greater procedural memory was required in the SRT task, learning in adults largely depended on offline learning; and, the age-related differences in learning generalization vanished, suggesting that offline learning may facilitate the generalization of implicit motor sequence learning. Taken together, results from these studies found two age-related learning processes (i.e., online and offline learning) that drive the fast implicit sequence acquisition and demonstrated that the age-related online and offline learning may lead to children a superior ability in the generalization of motor sequence learning. These results extend our understanding of the age-related development of implicit motor sequence learning and provide potential insights into the question of why childhood is an optimal period for learning.