We studied time estimation in patients with frontal damage (F) and alcoholic Korsakoff (K) patients in order to differentiate between the contributions of working memory and episodic memory to temporal cognition. In Experiment 1, F and K patients estimated time intervals between 10 and 120 s less accurately than matched normal and alcoholic control subjects. F patients were less accurate than K patients at short (< 1 min) time intervals whereas K patients increasingly underestimated durations as intervals grew longer. F patients overestimated short intervals in inverse proportion to their performance on the Wisconsin Card Sorting Test. As intervals grew longer, overestimation yielded to underestimation for F patients. Experiment 2 involved time estimation while counting at a subjective 1/s rate. F patients' subjective tempo, though relatively rapid, did not fully explain their overestimation of short intervals. In Experiment 3, participants produced predetermined time intervals by depressing a mouse key. K patients underproduced longer intervals. F patients produced comparably to normal participants, but were extremely variable. Findings suggest that both working memory and episodic memory play an individual role in temporal cognition. Turnover within a short-term working memory buffer provides a metric for temporal decisions. The depleted working memory that typically attends frontal dysfunction may result in quicker turnover, and this may inflate subjective duration. On the other hand, temporal estimation beyond 30 s requires episodic remembering, and this puts K patients at a disadvantage. (JINS, 2000, 6, 517–528.)