In observed conversations auditory and visual cues are used to shift attention towards the current speaker. The aim of the present study was to investigate whether one sensory modality dominates this cueing in prelinguistic and linguistic children and adults. To this end, participants from three age groups (1-year-olds, 3-year-olds and adults) were presented with three videos of conversations between two puppets: one condition with speech starting before the mouths moved (auditory first), one with the mouths moving before speech was audible (visual first) and one synchronous condition. Sound and vision were asynchronous only at the onset of each turn, at the end of turns, speech and mouth movements were always synchronous. We measured eye movements and calculated the latency of gaze shifts after the onset of speech and / or movement of the mouth at the beginning of turns. All age groups showed shorter latencies when auditory and visual cues occurred synchronously. The 1-year-olds and adults showed no difference between auditory-first and visual-first conditions, whereas the 3-year-olds showed significantly shorter latencies in the visual-first than auditory-first condition. These results suggest that already prelinguistic children use integrated auditory and visual cues in observed conversations to shift their attention between speakers. Furthermore, they indicate developmental differences: 3-year-olds rely more on visual than auditory cues. It has been shown in previous studies that 3-year-olds focus on different aspects of conversations than younger children and adults (e.g. intonation). It is possible that they also use visual cues more to facilitate the following of conversations.