The present invention provides a mask alignment mechanism which reduces the occurrence of particles and which aligns a mask with high accuracy, and a vacuum processing apparatus including such a mask alignment mechanism. A mask alignment mechanism according to one embodiment of the present invention includes a substrate holder which is movable up and down when a substrate is transferred and on which four taper pins are formed, and a mask in which grooves are formed. The taper pins can be inserted into the grooves, respectively. The taper pins include a pair of long taper pins and a pair of short taper pins. The taper pins in each pair are disposed to face each other across the substrate. Tapered surfaces formed in the long taper pins and tapered surfaces formed in the short taper pins are located at different heights.