The purpose of this study was to examine the cognitive and neuropsychological constructs that are conceptually related to poor insight in obsessive-compulsive disorder (OCD). The relationship between dimensions of insight (Brown Assessment of Beliefs Scale; BABS) and cognitive (magical thinking, paranoia/suspiciousness), metacognitive (metacognition, decentering, cognitive flexibility), and neuroopsychological indices of cognitive flexiblity were examined. Participants with OCD (N = 80) referred for treatment at an outpatient anxiety disorders clinic completed a clinical interview, a brief battery of neuropsychological measures, and a computer-administered questionnaire package assessing the variables of interest. Lower metacognition (i.e., Beck Cognitive Insight Scale [BCIS], composite score) was significantly associated with poorer insight (BABS total; ρ = -.38), and Metacognitions Questionnaire-30 cognitive self-consciousness subscale was negatively correlated with insight regarding a psychiatric source for one’s symptoms (ρ = -.24). Stroop interference was the only neuropsychological variable associated with BABS total score (ρ = -.23), but was not a unique predictor of insight in a regression with BCIS composite scores predicting insight. Nearly all of the variance in insight was accounted for by BCIS composite scores (R = .43, R2 = .18), indicating that metacognition, but not cognitive flexibility, contributes most strongly to clinical insight. Finally, insight decreased when OCD symptoms were activated for both the good and poor insight groups, F(1,78) = 119.29, p < .001, partial η2 = .61, and did not significantly vary as a function of insight group status, F(1, 78) = 3.24, p = .08, partial η2 = .04. Implications, limitations, and directions for future research are discussed.