In job rotation programs, higher level managers or human resource specialists often assign multiple leaders to an employee following a random sequence, because they have no method or no prior knowledge about this decision. In this paper, we focus on a destructive leader experience— working with an abusive leader. Random leader sequence in job rotation implies that some employees are assigned with an abusive supervisor early on while others are assigned with one late in the process. It thus creates an interesting theoretical tension: is it better off to have primacy-based leader sequence (i.e., abusive supervisors should appear early in the sequence) or recency-based leader sequence (i.e., abusive supervisors should appear late in the sequence)? Based on a 4-month job rotation program that involves 291 nurses and 513 nurse supervisors, we found that a primacy-based abusive supervision sequence consistently has a positive effect on both proximal rotation outcome (i.e., qualified exam performance and job performance) and distal rotation outcomes (i.e., job performance and bonus). The model not only adds a new inquiry of the leader sequence problem to leadership and job rotation research, but also offers much needed action implications to managers about how to assign leaders in job rotation.