Abstract BACKGROUND Alongside with the development of hand held equipment, echo is becoming more accessible to nontraditional settings. General and family medicine (GPs) are at the forefront of any health system, but the use of cardiac echo by GPs is still unexplored. PURPOSE: This study aims to evaluate the accuracy of echocardiography assessment performed as an extension of the physical exam by GPs. METHODS: Two GPs underwent standard echocardiography training for 6 months. Subsequently, patients with diabetes or arterial hypertension were submitted to an echocardiogram performed by their GPs using a pocket ultrasound. Views obtained (with and without color Doppler) were parasternal long and short axis, apical and subcostal, with measurements of the posterior wall (PW), interventricular septum (IVS), left ventricle (LV), left atrium (LA), aorta and inferior vena cava (IVC). Studies were reviewed at the workstation and measured by two experts who classified the exams according to the image quality. RESULTS: Sixty echocardiograms were analyzed (mean age of 61y, 42% females and 58% males). In 50% the image quality was considered sufficient, 42% considered good and 8% considered bad. There were statistically significant differences between the GPs and expert measurements on the sinus of Valsalva (30.3 ± 3 to 28.8 ± 3.1mm, p = 0.001), LV in systole (p < 0.001, 28[24,31] for 31[29,34]mm), IVS (p = 0.001, 10.9 ± 1.7 for 10.1 ± 1.5mm), PW (p = 0.018, 8.7 ± 1.1 to 9.2 ± 1.6mm) and TAPSE (p = 0.021, 20.9 ± 2.3 to 20.1 ± 2.7mm). There were no significate differences in the measurements of the ascending aorta, LA, LV in diastole and IVC. Agreement between the GPs and the experts was moderate for the evaluation of LVH (k = 0.48). The concordance was substantial for evaluation of LV function (k = 0.66) and excellent for evaluation of pericardial effusion (k = 1) and right ventricular function (k = 1). Concordance was excellent for mitral insufficiency (k = 0.83) and substantial for aortic insufficiency (k = 0.68) and tricuspid insufficiency (k = 0.61). A case of mitral stenosis was identified by both. GPs signalized a case of mild aortic stenosis that the expert did not consider. CONCLUSION: GPs trained in echocardiography, using pocket ultrasound, can obtain cardiac images with sufficient quality for interpretation by experts in the majority of cases. In this study, differences in dimensions might be explained by interobserver variability and/or by the performance of measurements in different environments, mobile vs workstation. Overall the differences were minor and clinically meaningless.