fMRI at ultra-high field strengths (≥7T) allow submillimeter voxels across different cortical layers. A big challenge to infer layer-dependent activity information, however, is to find an fMRI contrast that has the best combination of local specificity to microvascular responses within cortical layers and sensitivity to detect activity changes. Here we compare contrast mechanisms that have been proposed to be applicable to layer-dependent fMRI, including blood volume fMRI with VASO, SE-BOLD, T2/T1ρ-prep-BOLD, diffusion weighted T2-prep-BOLD. We find that CBV-weighted VASO and T2-prep methods have a favorable compromise between sensitivity and specificity. Hence, we believe that these fMRI methods might be better suited for layer-dependent applications than conventionally used GE-BOLD and SE-EPI.