We use a complete and rigorous statistical indicator to measure the level of concordance between cosmological data sets, without relying on the inspection of the marginal posterior distribution of some selected parameters. We apply this test to state of the art cosmological data sets, to assess their agreement within the ΛCDM model. We find that there is a good level of concordance between all the experiments with one noticeable exception. There is substantial evidence of tension between the CMB temperature and polarization measurements of the Planck satellite and the data from the CFHTLenS weak lensing survey even when applying ultra conservative cuts. These results robustly point toward the possibility of having unaccounted systematic effects in the data, an incomplete modelling of the cosmological predictions or hints toward new physical phenomena.Our present understanding of the universe is based on the combination of several different cosmological observations that are joined in order to exploit their complementary sensitivities to distinct characteristics of our universe. Several supernovae (SN) surveys are added together, in a single catalogue, to measure the expansion history at late times. Different Baryon Acoustic Oscillations (BAO) surveys provide independent measurements of a cosmological standard ruler at several times. Large scale structure and weak lensing surveys measure the correlation of galaxies and weak lensing shear in many different redshift bins. These are combined together to get tomographic information on the clustering of cosmic structures. At last, measurements of the Cosmic Microwave Background (CMB) are reaching an extraordinary level of sensitivity, that allows to measure the CMB temperature fluctuations along with CMB polarization and lensing. These data are then joined together to exploit CMB sensitivity to both early and late times cosmology. In the future, cosmological studies are going further in this direction. Wide large scale structure surveys, like Euclid [1], will combine maps of galaxies at several different redshifts, that will be joined with measurements of the CMB from the Planck satellite [2] and sub-orbital experiments. The observational efforts, that are driving cosmology toward a phase of extremely accurate, large scale, measurements, will all be joined together to learn all possible information about the initial conditions and the evolution of our universe. In this program, however, a problem arises. How can we be sure that the data sets, that we will be collecting, form a coherent picture, when interpreted within a model? How do we quantify the agreement between them, to be aware of the possible presence of unaccounted systematic effects or hints toward new physical phenomena? Testing the agreement between data sets, in a rigorous way, that goes beyond the comparison of the marginal distribution of some parameters, is critical in answering these questions. The posterior of the model parameters is, in fact, not guaranteed to show tensions due to the marginalizatio...