Miller, Maples, and Campbell (this issue) present evidence that Rosenthal and Hooley’s (2010) concerns regarding the Narcissistic Personality Inventory’s (NPI) relation to psychological health may be unwarranted. To resolve this issue empirically, we conducted a meta-analysis (k=54, N=38,932). Meta-analytic results revealed that a subset of NPI items were indeed problematic; items that function poorly at differentiating narcissists from non-narcissists accounted entirely for the NPI’s connection to psychological health. These items were also strongly associated with self-esteem, but unrelated to aggression/anger. In contrast, the remaining NPI items were unrelated to psychological health, but associated with aggression/anger. We conclude that although the NPI measures narcissism, its poorest functioning items also link it to outcomes unrelated to narcissism.