We perform a Bayesian model selection analysis in the the R-parity conserving MSSM to compare two different assumptions: whether the lightest neutralinos make all or only part of the cold dark matter. This corresponds to either imposing full WMAP relic density limits or just its upper bound for constraining the MSSM parameters. We consider several realisations of the MSSM, namely, three GUT-scale SUSY breaking scenarios with a handful of parameters corresponding to the CMSSM, anomaly mediation and the large volume string scenarios as well as the weak-scale 25-parameter phenomenological MSSM (pMSSM). The results give a data-based quantitative evidence for a multicomponent cold dark matter. The pMSSM posterior samples indicate that the choice of imposing full WMAP limits or just its upper bound affects mostly the gaugino-higgsino content of the neutralino and, against naive expectations, essentially not any other sector.

version to appear in Physcis Letters B