In this work we have analyzed changes in the heart rate variability (HRV) during exercise test comparing them with the ST deviation criteria to improve the diagnostic value of the exercise test. Coronary angiography was considered as gold standard to establish the classification of patients in two groups (ischemic and non-ischemic). ST deviations and HRV indexes were automatically measured and used as independent factors in discriminant analysis to find those more useful to classify both groups. Several approaches were performed starting with different sets of variables. Results showed that by using only the ST indexes it is possible to correctly classify 76% of patients. The inclusion of HRV indexes improves the exactness up to 84%. The very high frequency (0.4 to 1 Hz) at the stress peak has shown to have diagnostic value. Adding the age and the maximum heart rate the exactness goes up to 87.4% (sens. 85.5%, spec. 89.1%), close to that obtained by exercise echocardiography or exercise nuclear imaging.