Hindsight bias has recently been conceived of not as a unitary phenomenon but as a conglomerate of 3 separate phenomenological manifestations ("hindsight components"; Blank, Nestler, von Collani, & Fischer, 2008): memory distortions, impressions of foreseeability, and impressions of inevitability. These components are thought to be fundamentally different in nature, to be influenced by different processes, and to serve different functions. This article provides strong evidence for the separate components view and its underlying assumptions by demonstrating theoretically predicted dissociations between the components. In Experiment 1, for example, we used a memory encoding manipulation to specifically influence the amount of hindsight memory distortion but not participants' inevitability impressions. Conversely, varying the number of provided reasons for an event outcome affected inevitability impressions but left memory distortion untouched. Similar results--using different theoretically derived manipulations--were obtained between foreseeability impressions and memory distortions (Experiment 2) and between inevitability and foreseeability impressions (Experiment 3). Theoretical and practical consequences of these results and of the separate components view are discussed. (Contains 1 table, 2 figures and 8 footnotes.)