Incremental exercise with blood sampling to characterise the lactate-workload relationship is commonplace in exercise physiology laboratories to determine exercise intensity and monitor training adaptation. Various lactate markers exist, yet the reproducibility of speed at fixed and modelled plasma lactate markers remains unclear during incremental treadmill running of different stage durations. Twenty-one healthy volunteers (male = 15; age 22±3 yr, height 175±8 cm, body mass 70±11 kg) performed repeat running trials of 4- and 8-min stages on a powered treadmill at a 1% gradient. Trials commenced at speeds of 1.94 and 2.22 m.s-1 for females and males, respectively, with increments of 0.28 m.s-1. Duplicate fingertip capillary blood samples were drawn into non-lysed tubes at rest and during the final 30 s of each stage and analysed for plasma lactate (2300 STAT Plus™, YSI Life Sciences, US). Lactate analysis software (1) employing 3rd degree polynomial fitting (r2 = 0.983 ± 0.030) was used to determine running speed at the fixed markers: 2.0, 3.5, and 4.0 mmol.L-1, the 1 mmol.L-1 rise from baseline, and the modelled markers: deviation maximum (Dmax), lactate threshold (LT) and log-log LT. Dmax was the speed at the maximum perpendicular from a line connecting the first and the final lactate-speed points to the polynomial. LT method employs a ‘broken stick’ model (1) identifying the dividing point between two fitted regression lines as the corresponding speed, with log-log LT applying a log transformation (2). Reproducibility was assessed using Pearson’s correlation coefficients and limits of agreement (LoA) (3). With the exception of Dmax for 8-min stages (P<0.05), marker running speed for repeat trials of 4- and 8-min stages were similar. Correlation coefficients for marker running speed between repeat trials of 4- and 8-min stages were: r = 0.57, 0.77 (2.0 mmol.L-1); 0.58, 0.75 (3.5 mmol.L-1); 0.65, 0.85 (4.0 mmol.L-1); 0.70, 0.94 (1 mmol.L-1 rise); 0.45, 0.62 (Dmax); 0.57, 0.61 (LT) and 0.10, 0.43 (log-log LT), respectively. For 4-min stages, the LT recorded the lowest mean difference between trials (0.04 m.s-1), yet the 1 mmol.L-1 rise had the narrowest LoA (-0.84 to 0.71 m.s-1). For 8-min stages, the 1 mmol.L-1 rise demonstrated the lowest mean difference between trials (-0.05 m.s-1) and the narrowest LoA (-0.35 to 0.26 m.s-1). Agreement was greater for all lactate markers during 8-min stages, yet irrespective of stage duration, Dmax and log-log LT were not reproducible. These findings suggest greater reproducibility of speed at fixed lactate markers as opposed to curve-modelled markers (with the exception of LT) for incremental treadmill running of 4- and 8-min stages. The application of summary markers from lactate-speed data for intensity prescription of treadmill running should take into account marker suitability and stage duration.