To develop a clinical score to determine preclinical predictors of systolic dysfunction in an outpatient elderly population without a diagnosis of heart failure (HF). PULSE-HF is a cross-sectional study in elderly at-risk (coronary artery disease, diabetes or hypertension) outpatients without a diagnosis of heart failure (HF). The objective in this population was to develop a clinical score to determine preclinical predictors of systolic dysfunction. Clinical and geriatric variables were analyzed; independent predictive factors in the logistic regression analysis were included for the score calculation. Of the 722 subjects enrolled, 47 (6.5%) had a left ventricular ejection fraction (LVEF) < 50%, and 15 (2.1%) a LVEF < 40%. Mean age was 76.5 years (5.18) and 445 (61.6%) were female. Multiple logistic regression analysis identified abnormal Q waves (odds ratio OR: 4.36; P = 0.003), cardiomegaly (OR: 3.32; P < 0.001), right bundle branch block (OR: 2.84; P = 0.011), cognitive dysfunction (OR: 2.14; P = 0.027) and NT-proBNP (OR 5.43; P < 0.001) as independent predictors of LVEF < 50%. Two prediction scores were built, without and with NT-proBNP inclusion; the area under ROC curves were 0.70 and 0.76, respectively. As the score increased, the sensitivity decreases but increases specificity, and accuracy (97.17% and 91.64% respectively in ≥6 points). NT-proBNP was associated with an increment in the performance (accuracy of 93.18% for score ≥10). We conclude that a simple score using clinical information might be useful to predicting asymptomatic systolic dysfunction in the elderly.