McDaniel and Bugg (2008) proposed that relatively uncommon stimuli and encoding tasks encourage elaborative encoding of individual items (item-specific processing), whereas relatively typical or common encoding tasks encourage encoding of associations among list items (relational processing). It is this relational processing that is thought to result in better memory for the serial order of a study list. We report 4 experiments examining memory for order demonstrating that (a) both semantic and orthographic tasks can impair memory for order when performed on individual items, (b) item-specific processing is not necessary for impairment because even an item-generic key press task harms memory for order, (c) impaired memory for order is due primarily to distraction during the processing of an item rather than between list items, and (d) even an unusual processing task will preserve memory for order as long as that task encourages the encoding of item-to-item relations. These findings suggest that an encoding task will disrupt order memory only when it is both attention grabbing (either through its atypicality or by requiring an overt response) and nonrelational.