We have devised a new technique for identifying stellar mass black holes in low mass X-ray binaries, and have applied it to XMM-Newton observations of two X-ray sources in M31. In particular we search for low accretion rate power density spectra; these are very similar for all LMXB, whether the primary is a black hole or a neutron star. Galactic neutron star LMXB exhibit these distinctive PDS at very low luminosities (~ 1036 erg s–1) while black hole LMXB can exhibit them at luminosities > 1038 erg s–1! Following the work of van der Klis (1994), we assume a maximum accretion rate (as a fraction of the Eddington limit) for low accretion rate PDS that is constant for all LMXB, and obtain an empirical value of ~10% Eddington. We have so far discovered two candidate black hole binaries in M31, exhibiting low accretion rate PDS at up to 3 x 1038 and 5 x 1037 erg s–1. If we assume that they are at <10% Eddington, they have minimum masses of 20 and 4 M⊙ respectively. Furthermore, any LMXB exhibiting a low accretion rate power density spectrum at a luminosity > 5 x 1037 erg s–1 is likely to have a black hole primary.