Traditionally, the traffic levels observed within a zone (here, one of the 21 French administrative regions) are explained by the characteristics of its inhabitants: actual income, car ownership rate, etc. But by definition, traffic (here, private vehicles) involves exchanges between regions whose distances from each other vary. Spatial econometrics with panel data serves to highlight these interactions and, via regional series dating back to 1985, provides new estimations of the elasticities of private vehicle traffic, expressed as the average annual distance travelled per vehicle, to fuel price, to actual income and to ownership rate. Several spatial specifications are used, namely the autoregressive spatial models with random and fixed effects. They are estimated via a maximum likelihood estimator based on a two-step iterative procedure. The results reveal significant spatial interdependences characterising regional private-vehicle traffic. The advantage of introducing the spatial dimension is that it reduces any potential specification biases that might have a significant effect on the elasticities estimated. Furthermore, this type of model allows for an estimation of the direct, indirect and total effects associated with fuel price, actual income and ownership rate. Globally, the bounds of the 95% confidence intervals for the direct elasticities to fuel price, ownership rate and actual income are respectively [- 0.11; - 0.07], [- 0.36; - 0.23] and [0.38; 0.50]. The second-car boom brought down the use of each vehicle, hence a negative elasticity to the ownership rate, which has gradually picked up over time.