Context. We report on the results of the first XMM-Newton systematic "excess variance" study of all the radio quiet, X-ray un-obscured AGN. The entire sample consist of 161 sources observed by XMM-Newton for more than 10 ks in pointed observations, which is the largest sample used so far to study AGN X-ray variability on time scales less than a day. Aims. Recently it has been suggested that the same engine might be at work in the core of every black hole (BH) accreting object. In this hypothesis, the same variability should be observed in all AGN, once rescaled by the M BH (M BH ) and accretion rate (ṁ). Methods. We systematically compute the excess variance for all AGN, on different time-scales (10, 20, 40 and 80 ks) and in different energy bands (0.3-0.7, 0.7-2 and 2-10 keV). Results. We observe a highly significant and tight (∼0.7 dex) correlation between σ 2 rms and M BH . The subsample of reverberation mapped AGN shows an even smaller scatter (only a factor of 2-3) comparable to the one induced by the M BH uncertainties. This implies that X-ray variability can be used as an accurate tool to measure M BH and this method is more accurate than the ones based on single epoch optical spectra. This allows us to measure M BH for 65 AGN and estimate lower limits for the remaining 96 AGN. On the other hand, the σ 2 rms vs. accretion rate dependence is weaker than expected based on the PSD break frequency scaling. This strongly suggests that both the PSD high frequency break and the normalisation depend on accretion rate in such a way that they almost completely counterbalance each other (PSD amp ∝ṁ −0.8 ). A highly significant correlation between σ 2 rms and 2-10 keV spectral index is observed. The highly significant correlations between σ 2 rms and both the L Bol and the FWHM Hβ are consistent with being just by-products of the σ 2 rms vs. M BH relation. The soft and medium σ 2 rms is very well correlated with the hard σ 2 rms , with no deviations from a linear one to one correlation. This suggests that the additional soft components (i.e. soft excess, warm absorber) add a minor contribution to the total variability. Once the variability is rescaled for M BH andṁ, no significant difference between narrow-line and broad-line Seyfert 1 is observed. Conclusions. The results are in agreement with a picture where, to first approximation, all local AGN have the same variability properties once rescaled for M BH andṁ.