Four experiments examined the role of verbal processing at different stages of face recognition memory. In Experiment 1 participants learned faces with or without articulatory suppression, then engaged in an old/new recognition task. Using the same procedure, Experiment 2 examined performance under single and dual encoding conditions, using articulatory suppression and face verbalisation. In Experiment 3 performance deriving from these conditions was compared with a tapping control. The results were consistent; articulatory suppression impaired performance in comparison to the other conditions, which themselves did not differ. Experiment 4 examined the effects of postencoding verbalisation on performance, and showed some evidence for a standard verbal overshadowing effect. These results suggest that the role of verbal processing in face memory is complex, depending on the time when such processes occur. The results are discussed with reference to theories of verbal overshadowing.