Bayesian bounds incorporate prior knowledge on parameters of interest. Nonlocal bounds can provide more accurate prediction of the performance of estimators over the full range of possible mean-squared errors. For example, local bounds, such as the Cramer-Rao bound (CRB), provide especially inaccurate predictions under low signal-to-clutter-plus-noise ratio (SCNR) conditions. In this paper, we derive the Ziv-Zakai bound (ZZB) for joint location and velocity estimation for noncoherent, multiple-input multiple-output (MIMO) radar employing orthogonal waveforms for widely spaced antennas and white Gaussian clutter-plus-noise. The ZZB is a non-local Bayesian bound. We show that the ZZB is a comprehensive metric that captures the effect of the SCNR, the waveforms, and the other parameters of the radar system. The ZZB is shown to display three SCNR operating regions, namely the clutter-plus-noise, ambiguity, and asymptotic regions. The effects of different system configurations are explored through numerical studies.