A heuristic model of the relation between blood lactate (L) and VO2 during exercise was used to assess the comparative merits of two methods of reference threshold determination: the habitual linear interpolation method (LT), and a new method using inverse parabolic (second degree) interpolation (PT); the new method capitalizes on the demonstrated curvilinear relation between blood lactate and O2 uptake. Both types of interpolated thresholds were computed, and their error evaluated against the "true" 4-mmol lactate threshold. A combined parametric and Monte Carlo investigation showed that parabolic thresholds are generally superior, being less biased and more precise than their linear counterpart.