The Authors propose a computerized method in order to automatically detect the anaerobic threshold by the analysis of ventilatory parameters (VE, VO2). The algorithm calculates all possible linear fits and the relative standard error of the relationship between VE and VO2 beginning from the first set of 4 data (excluding the first 2 min of exercise) and increasing of 1 pair of data until peak exercise. Subsequently the program chooses the line that fits the greatest number of data with the smallest error. The ventilatory anaerobic threshold (SA Ve) is then calculated as the point at which the relationship between VE and VO2 is no more linear (i.e. when the VE measured gets over of 2 standard errors the calculated value on the basis of the regression formula). During the first phase of the study the method was validated against invasive AT determination by arterial lactate concentration (SA La) in 14 patients (7 athletes, 7 healthy sedentary subjects) during a symptom-limited ergospirometric test (in supine position, 10 W/min until exhaustion). Subsequently we studied the method repeatability in 20 men (10 normals, 10 patients with congestive heart failure who performed 2 ergospirometric tests on separate days. The results showed a good correlation when comparing each other the VO2 (SA Ve 26.88 +/- 4.24, SA La 25.95 +/- 3.88 ml/kg/min; r = 0.88) or the onset time (SA Ve 11.8 +/- 2.42, SA La 11.61 +/- 1.8 min; r = 0.91) of anaerobic threshold determined by the 2 methods.(ABSTRACT TRUNCATED AT 250 WORDS)