ABSTRACTthe piecewise multiphase regression method allows to flexibly combine linear and non-linear functions between a response and a predictor variable in a unique model, respecting the underlying physiology. routine data of exercise tests of 30 subjects including lactate, heart rate and Borg scale measurements between 80 and 320 Watt were evaluated by means of multiphase regression models in order to determine deflection points and linearity of curve segments. A two-phase linear-quadratic model was suited to define break points of lactate, heart rate and Borg scale versus workload curves. Whereas deflection points of the response variables heart and lactate corresponded largely to the aerobic threshold, deflection point of the Borg scale curve reflected high exercise intensity with lactate levels of about 6 mmol/l, i. e. above the anaerobic threshold. In conclusion, piecewise multiphase regression methods may be suited to improve interpretation of physical exercise tests with regard to lactate, heart rate or even Borg scales.