We developed a new diagnostic tool for predicting the severity of ankle sprains just after injury. Since hard data obtained by diagnostic imaging techniques are still imperfect, we decided to use data from individual medical history and signs and symptoms that are part of the admission routine. During a three month-period data were collected on thirty-five patients with lateral ankle sprains who visited the first aid department of the University Hospital of Maastricht. Assessments took place at admission and at two and four weeks after injury. Assessors were the first-aid physician, a physiotherapist and the patient. Dependent variables were healed ankle in two and four weeks. Predicting variables were the data obtained at admission by the physician, the physiotherapist and the patient. The ability to predict outcome after two and four weeks was determined in a bivariate analysis, followed by logistic modelling. Accurate prediction of recovery time at admission appeared to be possible. Best two weeks predictor was the modified function score, an accuracy of 97% was achieved. Four weeks prediction was most accurate when function score was used together with the report mark from the doctor and the palpation score (accuracy of 81%).