The purpose of this study was to determine the validity and reliability of a new mobile bike-powermeter, Polar S710, in laboratory and field conditions, against the SRM crankset. Eight trained subjects performed in a random order six uphill cycling trials of 6-min duration at three different intensities (60 %, 75 % and 90 % of peak power output [PPO]). In addition, 44 other cyclists performed in the laboratory three cycling bouts each of 5-min duration at three different pedal cadences (60, 90 and 110 rpm) at the same absolute intensity (approximately 150 W). Bias between the two devices was correlated (r = 0.79) with the mean power in field conditions; with the S710 reading higher (p < 0.001) by 7.4 +/- 5.1 % than the SRM in the range of power studied. In other words, the mean differences between the two devices increased as the exercise intensity increased. The mean power output obtained with S710 was significantly higher (p < 0.001) by 6.8 +/- 7.9 W (bias x divided-by random error = 1.042 x divided-by 1.049) than the power obtained with SRM in laboratory conditions. Ninety-five percent of the differences of power measured with the S710 ranged between 21.4 W above to 8.3 W below the SRM in laboratory conditions. Mean differences between the two devices increased as the pedalling cadence increased (0.6 +/- 3.8 %, 4.4 +/- 3.7 % and 7.8 +/- 4.4 % at cadence of 60, 90 and 110 rpm respectively). Coefficients of variation in mean power across the four field-based trials at 75 % PPO was 2.2 % and 1.9 % for S710 and SRM, respectively. In conclusion, the S710 recorded power outputs higher than the SRM system in both field and laboratory conditions. Pedalling cadence and exercise intensity influenced differences in mean power. These characteristics make S710 a useful device for recreational cyclists but not for elite cyclists or scientists who require a greater accuracy and validity. However, the limits of the present study (short-term duration testing; single tested variables as intensity, posture, pedalling cadence) require further investigation for generalizing the present results to extensive use in "real world" cycling.