In the winter sport of luge successful start is a prerequisite for a good total run result. Athletes train their starting technique not only at luge tracks themselves, but also on specially built start ramps, both iced and for roller sleds. Lugers begin their movement along the track by pushing from the start grips and then continue accelerating by pushing off the ice surface with both hands several times before laying down on the sled. Before leaving the start grips athletes perform forth-and-back rocking motion, whilst holding the handles. Measurement of start interval time begins at a certain distance from the start grips, when athletes had already left the handles. The last preparation phase before taking-off the handles was analyzed in the study. During this phase athletes first move as far forward as possible (position “a”), then push themselves maximally backward (position “b”), and then again move forward, using their arms and body force to achieve maximal acceleration, and finally leave the handles (position “c”). Data were collected during start trainings at a specially built artificial skeleton-luge-bobsleigh iced start ramp, and compared to data achieved from competition at National Federation level that took place at the same ramp after the training camp. Three junior athletes (one male and two female) performed 4 starts during the competition, these starts and first 4 out of 10 starts during the training were analyzed. Total start time, time at which athletes changed their position from “a” to “b”, and time from position “b” to “c” were measured. Total start time was measured with accuracy up to 1/100th of second using photocells. Other times were detected from video records performed with Casio EX-FH20 digital camera at 210Hz recording frequency. Angle between right thigh and torso at position “a” (angle α), torso inclination angle to horizontal line (angle β1) and right thigh inclination angle to horizontal line (angle β2) at position “b”, and maximal angle between right thigh and torso when moving from position “b” to “c”(angle γ) were also estimated from video records. Start time for male athlete (athlete C) was significantly shorter during competition, than during training (2,16 s (±0,01) and 2,19 s (±0,01) respectively). Start times did not differ significantly between female athletes and were similar both for competitive and training events (2,39 s (±0,01) and 2,37 s (±0,02) respectively for athlete A, and 2,39 s (±0,01) in both cases for athlete B). Despite differences in the start time between male and female athletes, time from position “b” to “c” showed to be similar for all athletes both in training and competitive events. Athlete C and athlete A were significantly faster in changing position from “a” to “b” during competition than during training; for female athlete this time improvement was accompanied with decrease in angle α and increase of angle β1, and opposite was the case for male athlete. All athletes tended to have angles γ larger during competition than during training. Athlete A, who tended to be slower during competition than during training, also showed a trend to be less stable in measured angles during competition if compared to other athletes.