It is shown that the errors in the absolute values of the measured delay times (after correction for diffraction phase and bond thickness) in the pulse‐echo overlap and the pulse‐superposition methods can be less than 0.015τ and 0.027τ, respectively, where τ is the period of the fundamental radio frequency of the transducer used. If broad‐band pulses are used with the pulse‐echo overlap method, then the error is larger but still less than 0.10τcat the center frequency. Translated into fractional errors in the single round‐trip delay time, the errors may be as small as a few parts per million. It is emphasized that the diffraction phase can introduce an error up to 0.25τ if the diffraction phase correction is not applied. Empirical methods for finding some contributions to the total error are given.