Retrieval of target information can cause forgetting for related, but non-retrieved, information – retrieval-induced forgetting (RIF). The aim of the current studies was to examine a key prediction of the inhibitory account of RIF – interference dependence – whereby ‘strong’ non-retrieved items are more likely to interfere during retrieval and therefore, are more susceptible to RIF. Using visual objects allowed us to examine and contrast one index of item strength –object typicality, that is, how typical of its category an object is. Experiment 1 provided proof of concept for our variant of the recognition practice paradigm. Experiment 2 tested the prediction of the inhibitory account that the magnitude of RIF for natural visual objects would be dependent on item strength. Non-typical objects were more memorable overall than typical objects. We found that object memorability (as determined by typicality) influenced RIF with significant forgetting occurring for the memorable (non-typical), but not non-memorable (typical), objects. The current findings strongly support an inhibitory account of retrieval-induced forgetting.