Ultrasound velocity measurements in medicine and biology usually are performed using relatively small measurement chambers. When the pulse-echo method is used, the presence of the reflector close to the transducer can cause essential diffraction errors. These errors may be reduced using an additional buffer rod as a waveguide between the transducer and the measurement chamber. The objective of the presented work was analysis of diffraction errors in measurement chambers with a buffer rod. The work was performed in two steps. In the first stage propagation of transient ultrasonic waves in a buffer rod was analysed using an axisymmetric finite element model. This approach enables all dimensions of the measurement chamber and the waveguide to be taken into account, but is less accurate in the time domain. In the second step the absolute values of diffraction errors were evaluated using a mixed analytic-numeric disk shaped transducer diffraction model. In this case only the dimensions of the waveguide and measurement chamber along the wave propagation direction were taken into account. Diffraction errors were calculated by simulating small changes of ultrasound velocity in the liquid under investigation.