Until present, workload in cycling competitions was mostly described through heart rate measurements and indirect estimation of exercise intensity. However, the most accurate variable to describe workload in cycling is to measure directly the power applied by the rider to propel his bicycle. PURPOSE: To illustrate direct power output of professional cyclist during a three-week stage race (Tour de France). METHODS: 143 Stage recordings of 15 riders of the 2005 Tour de France were monitored using a mobile power measurement device (SRM Trainingssystem, Julich, Germany), which recorded Power (W) and cadence (rpm). The studied riders were not contending for a high overall ranking but represented a broad range of skills from mountain specialists to team helpers. According to their topographic characteristics, stages were categorized into flat (7 stages/54 recordings), hilly (5/43) and mountain stages (6/46). Data was electronically processed, overall averages for power in different stage types, time in different power- and cadence bands and maximal mean power (MMP) for various durations were calculated. RESULTS: Power output averaged 218 +/− 21 W (3.1 +/− 0.3 W/kg) for flat, 234 +/−13 W (3.3 +/− 0.2W/kg) for mountain-, and 228 +/− 22 W (3.3 +/− 0.3 W/kg) for hilly stages. Mean cadence was 74 +/− 6.3 rpm for flat, 69 +/− 3.2 rpm for mountain-, and 70 +/− 5.2 rpm for hilly stages. Riders spent significantly more race time at medium- and high power output (76 % vs. 66 %) and lower cadence (56 % vs. 36 %) in mountain and hilly stages compared to flat stages. Mountain stages showed higher MMP (426 W vs. 385 W) for longer durations (1800 sec) and lower MMP (1046 W vs. 1179 W) for shorter durations (15 sec) compared to flat stages. CONCLUSION: Measurements of power output allow a direct description of the workload sustained by an athlete during cycling competition. Our data shows that different topographical characteristics of a race require distinct power output capacities with long, medium- to high intensity efforts at low cadence during mountain stages and short, high intensity power bursts during flat stages. As the presented data was obtained from riders without ambitions for a high overall ranking in the three week race, it can be speculated that riders aiming for a top overall finish display even higher average power outputs during decisive stages.