A non-invasive ultrasonic (US) technique of tendon force measurement has been recently developed. It is based on the relationship demonstrated between the speed of sound (SOS) in a tendon and the traction force applied to it. The objectives of the present study were to evaluate the variability of this non-linear relationship among 7 equine superficial digital flexor (SDF) tendons, and the reproducibility of SOS measurements in these tendons over successive loading cycles and tests. Seven SDF tendons were equipped with an US probe (1MHz), secured in contact with the skin overlying the tendon metacarpal part. The tendons were submitted to a traction test consisting in 5 cycles of loading/unloading between 50 and 4050N. Four tendons out of the 7 were submitted to 5 additional cycles up to 5550N. The SOS-tendon force relationships appeared similar in shape, although large differences in SOS levels were observed among the tendons. Reproducibility between cycles was evaluated from the root mean square of the standard deviations (RMS-SD) of SOS values observed every 100N, and of force values every 2m/s. Reproducibility of SOS measurements revealed high between successive cycles: above 500N the RMS-SD was less than 2% of the corresponding traction force. Reproducibility between tests was lower, partly due to the experimental set-up; above 500N the difference between the two tests stayed nevertheless below 15% of the corresponding mean traction force. The reproducibility of the US technique here demonstrated in vitro has now to be confirmed in vivo.