We have reconstituted the fusion and assembly of vesiculated Golgi membranes (VGMs) into functionally active stacks of cisternae. A kinetic analysis of this assembly process revealed that highly dispersed VGMs of 60-90 nm diameter first fuse to form larger vesicles of 200-300 nm diameter that are clustered together. These vesicles then fuse to form tubular elements and short cisternae, which finally assemble into stacks of cisternae. We now provide evidence that the sequential stack formation from VGMs reflects two distinct fusion processes: the first event is N-ethyl-maleimide (NEM)-sensitive factor (NSF) dependent, and the second fusion event requires an NSF-like NEM-sensitive ATPase called p97. Interestingly, while the earliest steps in stack formation share some similarities with events catalyzing fusion of transport vesicles to its target membrane, neither GTP gamma S nor Rab-GDI, inhibitors of vesicular protein traffic, inhibit stack formation.