A growing body of research suggests the presence of an attentional component in the memory advantage of animate items. Most research on this animacy effect is focused on remembering, but its effects on forgetting are less well-researched. With the use of pupillometry, we investigated the attentional processes present in the selective remembering and directed forgetting of animate and inanimate items. More specifically, we investigated whether external and internal attention are affected by the animacy status of to-be-remembered and to-be-forgotten words in an item-method directed forgetting task with retro cues, followed by a recognition task for all previously presented items. Our behavioral results demonstrate the directed forgetting effect: accuracy on the recognition task was higher for to-be-remembered words than for to-be-forgotten words. Additionally, we found an advantage of animacy, with higher accuracy for animate words than for inanimate words, regardless of cue. The pupillometry results demonstrate differential internal attention for to-be-remembered and to-be-forgotten words during cueing, with larger pupil sizes for remember cues than for forget cues, but no effect of animacy. We did however find an animacy effect on external attention during stimulus presentation, with smaller pupil sizes for animate words than for inanimate words, which may reflect precision of encoding. This suggests that not internal attention, but external attention is influenced by animacy status, potentially explaining the memory advantage of animate items through the richness-of-encoding hypothesis.