Four masked priming experiments are reported investigating the effect of inserting an unrelated word between the masked prime and the target. When the intervening word is visible, identity priming is reduced to the level of one-letter-different form priming, but form priming is largely unaffected. However, when the intervening word is itself masked, form priming is eliminated altogether, while identity priming remains at the level observed with a visible intervenor. Normal priming effects are obtained when the order of the masked prime and the intervenor is reversed. It is suggested that identity priming consists of two effects, one that depends on the prime being adjacent to the target (a semantic effect), and one that is independent of adjacency (a form effect). (Contains 3 figures and 4 tables.)