For fMRI to be able to provide insight in the cortical circuitry, the spatial resolution and specificity of the fMRI signal need to be sufficient to visualize the microarchitecture of the cortex at the laminar and columnar level. The current spatial resolution of the fMRI signal however is too coarse to be able to reliably visualize the cortical microarchitecture. Cortical columns of ~1 mm have been observed with fMRI, but achieving higher (submillimeter) resolution is problematic, because the specificity of the activation is determined by the hemodynamic properties of the vascular bed. The conventional Gradient- Echo (GE) sequence used for BOLD-fMRI is sensitive to signal from veins and venules, and is strongest at the cortical surface, where draining vessels are located. This has limited the spatial specificity of the conventional fMRI signal to about 1 mm. By using the Spin-Echo (SE) fMRI signal instead, which is more sensitive to the capillary fraction, and less sensitive to veins, the specificity of the fMRI signal can be enhanced. Sequence optimization allowed us to further increase spatial specificity, and to achieve submillimeter spatial resolution. This resolution allows visualization of the cortical laminae, as shown in V1 in the anesthetized macaque (figure). The SE-BOLD signal was localized to layer IV/Duvernoy layer 3, with little activation in the upper cortical layers. The spatial resolution and specificity shown here allows determination of differences in laminar profiles depending on visual input. When motion and flicker stimuli were compared, the unequal laminar distribution of motion-direction selective laminae could be clearly discerned. Our results indicate that the point spread function for SE-fMRI is 0.5 mm or less, and is sufficient to observe differences in functional activation at laminar resolution.