We have studied a sample of ~3600 broad-line QSO from SDSS DR12 (0.1<~z<~4), cross-correlating it with 3XMM DR6 to get X-ray fluxes and luminosities (using FLIX to get fluxes and upper limits) and using the results from unWISE to get MIR fluxes (and upper limits).

As in previous studies, we find a flattening of the 2-10keV vs. 6um luminosity relationship for the most luminous objects (43<=log(L_X/cgs)<=47,43<=log(nu*L_nu,6um/cgs)<=47) with respect to that found for Sy1, which we show that continues within our luminous sample.

Using the bolometric luminosities (44<=log(L_bol/cgs)<=47) from the optical continuum, this flattening appears to be due entirely to the known decrease of the L_X-L_bol ratio with increasing L_bol, since there is no trend of the L_MIR-L_bol ratio with L_bol.

We will also discuss the implication of these results on models for the X-ray and MIR emission of QSOs.