Much of the training of competitive telemark skiers is performed as dry-land exercises. The specificity of these exercises is important for optimizing the training effect. Our aim here was to study the activation of the knee extensor musculature and knee angular displacement during competitive telemark skiing and during dry-land strength training exercises to determine the specificity of the latter. Specificity was analysed with respect to angular amplitude, angular velocity, muscle action and electromyographic (EMG) activity. Five male telemark skiers of national and international standard volunteered to participate in the study, which consisted of two parts: (1) skiing a telemark ski course and (2) specific dry-land strength training exercises for telemark skiing (telemark jumps and barbell squats). The angular displacement of the right knee joint was recorded with an electrogoniometer. A tape pressure sensor was used to measure pressure between the sole of the foot and the bottom of the right ski boot. Electromyographic activity in the right vastus lateralis was recorded with surface electrodes. The EMG activity recorded during maximum countermovement jumps was used to normalize the EMG activity during telemark skiing, telemark jumps and barbell squats. The results showed that knee angular displacement during telemark skiing and dry-land telemark jumps had four distinct phases: a flexion (F1) and extension (E1) phase during the thrust phase of the outside ski/leg in the turn/jump and a flexion (F2) and extension (E2) phase when the leg was on the inside of the turn/jump. The vastus lateralis muscle was activated during F1 and E1 in the thrust phase during telemark skiing and telemark jumps. The overall net knee angular amplitude was significantly greater (P < 0.05) for telemark jumps than for telemark skiing. Barbell squats showed a knee angular amplitude significantly greater than that in telemark skiing (P < 0.05). The mean knee angular velocity of the F1 and E1 phases during telemark skiing was about 0.47 rad x s(-1); during barbell squats, it was about 1.22 rad x s(-1). The angular velocity during telemark jumps was 2.34 and 1.59 rad x s(-1) in the F1 and E1 phase, respectively. The normalized activation level of the EMG bursts during telemark skiing, telemark jumps and barbell squats was 70-80%. In conclusion, the muscle action and level of activation in the vastus lateralis during the F1 and E1 phases were similar during telemark skiing and dry-land exercises. However, the dry-land exercises showed a larger knee extension and flexion amplitude and angular velocity compared with telemark skiing. It appears that an adjustment of knee angular velocity during barbell squats and an adjustment of knee angle amplitude during both telemark jumps and barbell squats will improve specificity during training.