This study evaluated the reproducibility of laboratory based 20-km time trials in well trained versus recreational cyclists. Eighteen cyclists (age = 34 +/- 8 yrs; body mass index = 23.1 +/- 2.2 kg/m (2); VO(2max) = 4.19 +/- 0.65 L/min) completed three 20-km time trials over a month on a Velotron cycle ergometer. Average power output (PO) (W), speed, and heart rate (HR) were significantly lower in the first time trial compared to the second and third time trial. The coefficients of variation (CV) between the second and third trial of the top eight performers for average PO, time to completion, and speed were 1.2 %, 0.6 %, 0.5 %, respectively, compared to 4.8 %, 2.0 %, and 2.3 % for the bottom ten. In addition, the average HR, VO(2), and percentage of VO(2max) were similar between trials. This study demonstrated that (1) a familiarization session improves the reliability of the measurements (i.e., average PO, time to completion and speed), and (2) the CV was much smaller for the best performers.