Vehicle-to-Vehicle (V2V) radio channels have been considered as non-stationary due to the dynamic scatterers in V2V environments. To characterize its stationarity is of great importance as it determines the time-variant V2V channel statistics. Therefore, we analyze the non-stationarity of V2V radio channels using two metrics: the correlation matrix distance (CMD) and the shadow fading correlation. The analysis is based on the measurements conducted in a suburban area of Finland at 5.3 GHz. We use the Aalto channel sounder and a 30 × 4 MIMO system. The power delay profile (PDP), CMD, and shadowing are estimated from the measured channel impulse responses. The equivalent stationarity distance is found to be around 6 m, and it is observed that regions with large delay spreads correspond to smaller stationarity regions. The shadow fading auto-correlation coefficient is found to follow an exponential decay model and the decorrelation distance is considered as the shadowing-based equivalent stationarity distance. Based on the comparison of the equivalent stationarity distances estimated by the two metrics, it is found that the CMD metric is more sensitive to the changes of direction-of-arrival/departure, whereas the shadowing metric takes only into account the power variations over time. We also note that the estimation of shadowing metric is close to the results from CMD. This shows that the shadow correlation can also be used to characterize the non-stationarity of radio channels.