The technique of speed skating is unique in comparison to other sports that require human propulsion. Skaters generate a forward velocity by pushing off sideways. The ideal push-off technique is a trade-off: a more sideward directed push-off facilitates power production, but a more forward directed push-off increases the transfers rate of push-off power into a forward velocity. The exact trade-off for the ideal push-off technique is unknown. Insight in the distribution of power in more than one direction and into the energy expenditure during different push-off techniques helps a speed skater in improving his or her push-off technique to the ideal push-off technique. This study quantifies the effect of different push-off techniques on the mechanical power, power distribution and energy expenditure. The study was limited to a group analysis of three push-off techniques: the small, self-chosen and wide push-off technique. A three-dimensional power balance model was used to calculate the mechanical power in forward, sideward and upward components during speed skating. This model was driven by velocity and acceleration data, obtained by two synchronized measurement systems, and estimations of air and ice friction coefficients. The acceleration was measured by an Xsens MTI (Xsens) device and the velocity was calculated by fusing position (Local Position Measurement (LPM), Inmotio) and acceleration measurements (Xsens). The subject specific air and ice coefficients were estimated with position measurements (LPM) during gliding experiments. The mechanical power results were limited to the average mechanical powers of one representative stroke-cycle of skater and technique. The energy expenditure was estimated with steady state heart rate measurements (Polar) during speed skating. This study proved a significant difference in forward power component, sideward power component and the total mechanical power between the push-off techniques studied, as well as the energy expenditure between the push-off techniques. The sideward power increased from small to self-chosen to wide push-off technique. In addition, this study showed that the change in total mechanical power was mainly due to the change in the sideward power component and that the energy expenditure in the self-chosen technique was the lowest. The relative mechanical efficiency, the ratio between total mechanical power and steady state heart rate, was significantly different for the three different push-off techniques. In addition, the relative mechanical efficiency increased from small to selfchosen to wide push-off technique. In summary, of the three push-off technique was the self-chosen push-off technique the most energy efficient push-off technique; this technique required the lowest energy expenditure for the required forward velocity. However, this technique was not the most mechanical efficient. The total mechanical power of the wide push-off technique was generated most mechanical efficient. This is because the more sideward push-off will make the push-off velocity less independent of the moving velocity and can be freely chosen to the optimal leg extension velocity, but introduces more ’wasted power’ in non-effective movements for the performance. The selected measurement method of this study can improve the current determination of the ideal push-off technique: analyzing the important trade-off between the mechanical efficiency and push-off orientation. This method be used to identify small changes between push-off technique in the mechanical efficiency and push-off orientation, and thus be used to improve speed skating performance. However, the measurement accuracy of this method requires to be improved further.