Recent studies have shown systematic choice-supportive memory for past choices, wherein people tend to overattribute positive features to options they chose and negative features to unchosen options (Mather & Johnson, 2000, Mather, Shafir, & Johnson, 2000). In contrast, the present experiments showed no choice-supportive memory bias for assigned options. Rather than having a general motivation to recall the chosen or the assigned option in a more positive light, people appear to be influenced by heuristics that vary with context: In recalling past choices, people expect the chosen option to contain more positive and fewer negative features than do its competitors. In recalling past assignments, in contrast, people expect the assigned option to be remembered better than the unassigned alternatives. This vividness heuristic leads to systematic misattribution of new features to unassigned alternatives, but not in a manner supportive of the assigned option. Some implications of these findings are discussed.