Objective The Mini Mental State Exam (MMSE) has enjoyed widespread use as a dementia severity staging instrument (Perneczky et al., 2006). More recently, the Montreal Cognitive Assessment (MoCA; Nasreddine, 2005) has been advanced as a potentially superior measure with enhanced sensitivity to Mild Cognitive Impairment (MCI). To the authors’ knowledge, there are no published guidelines for staging dementia severity with the MoCA. The aim of this study was to evaluate the utility of the MoCA for dementia severity staging. Method Participants (N = 162) were drawn from a diagnostically heterogeneous retrospective sample of referrals to a multidisciplinary memory clinic. Participants were categorized as MCI, mild dementia, or moderate dementia using the Quick Dementia Rating System (QDRS) sum of boxes score. Receiver operating characteristics of the MoCA were calculated using MATLAB and optimal cutpoints were determined using Youden’s Index. Results The MoCA demonstrated some utility in differentiating MCI from all severity dementia as defined by the QDRS, with an optimal cutpoint of 17 (AUC = .75). Cut points of 17 and 14 best separated MCI from mild dementia (AUC = .72) and mild from moderate dementia (AUC = .66), respectively. These cutpoints were associated with modest sensitivity (.50 - .53) and reasonable specificity (.76 - .87). Average diagnostic accuracy was 69.5%. Conclusions This study suggests that the MoCA has some utility for dementia severity staging. Future work should replicate these findings in other clinical cohorts. The use of the QDRS (an informant report measure) as the severity criterion is a significant limitation of the present study.