The maximum urban heat island often occurs few hours after sunset. This may be explained by the thermal inertia of the urban canopy which is often much higher than that of rural sites. The cooling rate is an indicator of urban thermal inertia computed from on-site measurement but is mainly used to dissociate thermal behaviour difference between urban and rural sites. This paper proposes a new method to better dissociate the thermal inertia properties between urban sites from air temperature measurement. The first part of our paper presents the method of computation, its results under different meteorological conditions which are then compared to the results obtained from the cooling rate calculation. Our method is based on the phase shift computation of temperature diurnal cycles between several urban stations and a reference rural station. Fifteen minutes data collected during four years from ten temperature stations are used. The stations network is located in Nantes, the 6th largest city of France with a total population in its metropolitan area of 590 000 inhabitants. The climate is western European oceanic and is characterized by a relatively mild summer. The phase shift was first calculated considering different meteorological situations. The results show that the sun radiation amount affects directly the phase shift difference values. The wind speed and direction also play a role on the results even if the influence is lower. In a second step, our indicator is compared to the average cooling rate after sunset, an usual indicator of thermal inertia. The results show that the phase shift better dissociates the stations than the cooling rate regarding to their thermal inertia properties. In a second step, the phase shift results are analysed in relation with geographical indicators (facade density, vegetation density, etc.) calculated from BDTopo®. The reference surface for the spatial analysis is defined by concentric circles of different sizes around each measurement station. Results of linear regressions show that our new thermal inertia indicator is well correlated to geographical parameters (R² > 0,5 - e.g. for aspect ratio). These results can be used to identify high thermal inertia zones, where the urban heat island is expected to occur during night-time.