Evidence for generalised and episode-specific strength in judgements of recency and frequency of novel faces was investigated in two experiments. Novel faces (created using an identikit program) were presented either once or three times in either of two study sets, separated by a 4 minute (Experiment 1) or 24 hours (Experiment 2) delay. In Experiment 2, the study-test phases were preceded by a familiarisation phase in which faces were presented either with or without an occupational label (some faces were not familiarised). In both experiments, judgements of recency were consistent with greater reliance on generalised strength (an amalgamation of recency and frequency of presentation), although there was some evidence for a contribution of episode-specific information. Familiarisation had opposite effects on judgements of recency and frequency, with familiarised faces judged as having been studied more frequently (three times), but less recently (yesterday) than unfamiliarised faces. The latter result is consistent with participants discounting generalised strength when episode-specific information is inadequate.