The lactate threshold (LT) represents the onset of a metabolic acidosis during graded exercise testing (GXT). It is a valuable measurement in clinical exercise testing and correlates well with endurance performance. Our purpose was to compare three LT detection methods, namely, Inspection (work rate at onset of a systematic increase in blood lactate concentration determined by inspection of blood lactate versus work rate plot), 0.5 mM (work rate which just precedes a rise in blood lactate concentration of >0.5 mM) and log-log (work rate at the intersection of two linear lines in plot of log lactate versus log work rate where the residual sum of squares for both lines added together is minimized). Fourteen subjects underwent cycle ergometer GXT with blood samples obtained at the end of each 3-min work rate increment and analysed for lactate concentration. The mean +/- 95% confidence limits of work rates at LT for the Inspection, 0.5 mM and log-log methods were 104.5 +/- 28.0, 103.2 +/- 28.1 and 105.1 +/- 27.3 W, respectively. Repeated-measures analysis of variance yielded a non-significant F ratio. The Bland-Altman bias +/- 95% limits of agreement for Inspection versus 0.5 mM, Inspection versus log-log and 0.5 mM versus log-log were 1.3 +/- 20.6, -0.6 +/- 12.5 and -1.9 +/- 20.5 W, respectively. The intraclass correlation coefficients for Inspection versus 0.5 mM, Inspection versus log-log and 0.5 mM versus log-log were 0.978, 0.992 and 0.977, respectively. The results of this study suggest that all three methods detect the LT at the same work rate.