Cognitive screening tests are frequently used in brain tumor clinics. The Mini Mental State Examination (MMSE) is the most commonly used, and the Montreal Cognitive Assessment (MoCA) is an alternative. This study compares the diagnostic accuracy of both screening tests. Fifty-eight patients with brain tumors were prospectively accrued and administered the MMSE and MoCA, 67% of who completed a comprehensive neuropsychological evaluation as a gold standard comparison. Quality of life and community integration were measured with the Functional Assessment of Cancer Therapy-Brain (FACT-Br) and Community Integration Questionnaire (CIQ), respectively. At the pre-defined cut-off scores, the MoCA had superior sensitivity (61.9% vs. 19.0%, P < 0.005) and the MMSE had superior specificity (94.4% vs. 55.6%, P < 0.017). The areas under the ROC curve for the MMSE (0.615, standard error = 0.091) and MoCA (0.606, standard error = 0.092) were poor, indicating that at no single cut-off score is either test both sensitive and specific. Neither the MMSE (ρ = 0.12; P < 0.444) nor MoCA (ρ = 0.24; P < 0.108) were significantly correlated with the FACT-Br. The MoCA was modestly correlated with the CIQ (ρ = 0.35; P < 0.017), but the MMSE was not (ρ = 0.14; P < 0.359). The MMSE has extremely poor sensitivity. Using this test in clinical practice, research, and clinical trials will result in failing to detect cognitive impairment in a substantial percentage of patients. The MoCA has superior sensitivity, and is better correlated with self reported measures of community integration, and therefore should be preferentially chosen in practice and clinical trials.