Several years ago P. Horava suggested, basing on power-counting arguments, that a renormalizable unitary theory of gravity can be constructed at the expense of abandoning Lorentz invariance. I will review different forms of the Horava's proposal and explain why the power-counting by itself is not sufficient to establish renormalizability. Then I will present a rigorous proof of perturbative renormalizability for the so-called projectable Horava gravity. The key element of the proof is the choice of a gauge which ensures the correct anisotropic scaling of the propagators and their uniform falloff at large frequencies and momenta. I will also comment on the difficulties of this approach when addressing the renormalizability of the non-projectable model.