We compute the probability densities of the effective neutrino masses mβ and mββ using the Kernel Density Estimate (KDE) approach applied to a distribution of points in the (mmin,mββ) and (mβ,mββ) planes, obtained using the available Probability Distribution Functions (PDFs) of the neutrino mixing angles and mass differences, with the additional constraints coming from cosmological data on the sum of the neutrino masses. We show that the reconstructed probability densities strongly depend on the assumed set of cosmological data: for ∑jmj≤0.68 eV at 95%CL a sensitive portion of the allowed values are already excluded by null results of experiments searching for mββ and mβ, whereas in the case ∑jmj≤0.23 eV at 95%CL the bulk of the probability densities are below the current bounds.