Introduction The evaluation of cognitive impairment in adulthood merits attention in societies in transition and especially in people with chronic diseases. Screening tools available for clinical practice and epidemiological studies have been designed in high-income but not in resource-constrained settings. The aim of this study was to assess the agreement and bias of three common tools used for screening of cognitive impairment in people with hypertension: the modified Mini-Mental State Examination (MMSE), the Montreal Cognitive Assessment (MoCA), and the Leganés Cognitive Test (LCT). Methods A cross-sectional study enrolling participants with hypertension from a semi-urban area in Peru was performed. The three screening tools for cognitive impairment were applied on three consecutive days. The prevalence of cognitive impairment was calculated for each test. Pearson's correlation coefficients, Bland-Altman plots, and Kappa statistics were used to assess agreement and bias between screening tools. Results We evaluated 139 participants, mean age 76.5 years (SD ± 6.9), 56.1% females. Cognitive impairment was found in 28.1% of individuals using LCT, 63.3% using MMSE, and 100% using MoCA. Correlation coefficients ranged from 0.501 between LCT and MoCA, to 0.698 between MMSE and MoCA. Bland-Altman plots confirmed bias between screening tests. The agreement between MMSE and LCT was 60.4%, between MMSE and MoCA was 63.3%, and between MoCA and LCT was 28.1%. Conclusions Three of the most commonly used screening tests to evaluate cognitive impairment showed major discrepancies in a resource-constrained setting, signaling towards a sorely need to develop and validate appropriate tools. © 2016 The Authors