Our multi-epoch survey of ~20 sq. deg. of the Canis Major overdensity has detected only 10 RR Lyrae stars (RRLS). We show that this number is consistent with the number expected from the Galactic halo and thick disk populations alone, leaving no excess that can be attributed to the dwarf spheroidal (dSph) galaxy that some authors have proposed as the origin of the CMa overdensity. If this galaxy resembles the dSph satellites of the Milky Way and of M31 and has the putative Mv~-14.5, our survey should have detected several tens of RRLS. Even if Mv<-12, the expected excess is >10, which is not observed. Either the old stellar population of this galaxy has unique properties or, as others have argued before, the CMa overdensity is produced by the thin and thick disk and spiral arm populations of the Milky Way and not by a collision with a dSph satellite galaxy.