This thesis is concerned with the phenomenon of implicit learning. Implicit learning occurs "when there is a performance increase on some task, without an associated increase in verbal knowledge about the causes of this performance increase" (Bright, 1993 - p9). In chapter 1, two theoretical interpretations of this type of learning are described. The first proposes that implicit learning reflects the operation of a unconscious learning system (e.g. Reber, 1989). The alternative episodic processing view (Whittlesea and Dorken, 1993) suggests that implicit learning occurs when there is an indeterminate relationship between the explicitly held knowledge acquired during training, and the way this knowledge is used in the test. Two main experimental findings reported in the chapters 2-5 allow for a choice between the two main accounts mentioned above. Firstly, two implicit learning tasks (invariance learning [McGeorge and Burton, 1990] and sequence learning [Nissen and Bullemer, 1987]) demonstrate that learning is dependent on active processing of training stimuli. Secondly, findings from the invariant learning task indicate that episodic knowledge, rather than an abstract rule, is acquired in this type of learning. Both these findings are consistent with the episodic processing account of implicit learning, and not the separate system view. Furthermore, a specific prediction of the episodic account is also confirmed by the data reported in this thesis. This prediction is that the processing demands of the training and test periods must be consistent for successful performance. This support for the episodic processing account of implicit learning is accompanied by a caveat in chapter 6. It is suggested that the episodic processing view is unsuitable for understanding the type of processes occurring in all implicit learning tasks. A possible resolution is offered, in the suggestion that broadening the theoretical scope to a more general consistency model may allow the wider experimental context of implicit learning to be explained.