Measurements are reported on the propagation velocity of ultrasound in muscle tissue. This appears to be dependent on fiber direction for various muscle types (in vitro). Typical results are 1550 m/s perpendicular and 1557 m/s parallel to fiber direction (20 degrees C). A model is drawn up to relate such data to the elastic properties of muscle. It is shown that the assumptions: linear elastic behavior, homogeneity, and anisotropy with cylindrical symmetry, do not harmonize with the experimental results. A more elaborate model is suggested. Contraction dependence of the ultrasound velocity was investigated on human biceps muscle (in vivo) and frog gastrognemius muscle (in vitro). The measured velocity increase in biceps muscle during contraction--up to 3 m/s--is attributed to a varying blood content of the muscle. In (in vitro) frog muscle contraction is not accompanied by a change in ultrasound velocity. This suggests that increased muscle stiffness is not appreciably reflected in ultrasound velocity.