A pattern recognition analysis of six cardiovascular reflex tests was performed with the aim to increase their diagnostic efficiency. A normal model was defined by SIMCA method using log-transformed values of 85 healthy subjects. The model was tested with data from 60 patients with chronic renal failure. Specificity resulted 94% and sensitivity 96%. 55% of the uremics were recognized affected by autonomic damage. The method allowed a better definition of autonomic function in those patients defined as "borderline" on conventional evaluation of the tests. Pattern recognition analysis of autonomic tests provides a global assessment of all the tests performed. Its use should be encouraged as a diagnostic tool.