benefits. These are boosted if the screening test is less influenced by biases (e.g education), has easy application and interpretation, and involve ecological items to make it feel as natural and less harmful as possible.The GPCOG is known to be one of the most recommended cognitive screening instruments for primary care use, translated and validated in different countries and languages. Our aim is to verify its efficacy in a Brazilian primary care sample. Methods: 91 elderly patients presenting memory complaints attended in a primary care unit in Sao Paulo and their informants were assessed. Exclusion criteria were: severe hearing or visual deficits, current diagnosis of depression or delirium, lack of a reliable informant. In the absence of a dementia diagnosis, interview about clinical features and CAMCOG, MMSE, BADL and IQCODE results blindly separated control and demented groups. All were assessed with GPCOG. Spearman correlation coefficient was calculated for correlations. Sensitivity, specificity and accuracy were calculated for GPCOG. Cutoff point was evaluated by ROC curve. Results: 47 controls (C) and 44 demented (D) patients were identified. Groups differed in age. Results showed good correlation between GPCOG-total and CAMCOG [(C): rho 1⁄4 0.636; (D): rho 1⁄4 0.670; both p <0.001], MEEM [(C): rho1⁄4 0.634; (D): rho1⁄4 0.558; both p<0.001]. GPCOG-patient and CAMCOG [(C): rho 1⁄4 0.657; (D): rho 1⁄4 0.728; both p <0.001], MEEM [(C): rho 1⁄4 0.739; (D): rho 1⁄4 0.575; both p <0.001]. GPCOG-informant and IQCODE [(C): rho 1⁄4 -0.552; (D): rho 1⁄4 -0,719; both p <0.001] and BADL [(C): rho 1⁄4 0.668; (D): rho 1⁄4 0.635; both p <0.001].The GPCOG-patient cut-off score 2 showed 75% sensitivity, 86.8% specificity, and 82.4% accuracy. For GPCOG-informant cut-off score 3 showed 86.4% sensitivity, 69.1% specificity, and 74.7% accuracy. The GPCOG-total cut-off score 7 showed 90.9% sensitivity, 78.4% specificity and 83.5% accuracy. Conclusions: GPCOG seems to be a good cognitive screening instrument for use in Brazilian primary care context.