In preparing a halftone phase shift mask blank, a metal and silicon-containing compound film serving as a halftone phase shift film is formed on a transparent substrate by a co-sputtering process including the steps of disposing a metal-containing target and a silicon target in a chamber, feeding sputtering gases into the chamber, and applying electric powers across both the targets at the same time. The sputtered region area of the metal-containing target is smaller than the sputtered region area of the silicon target.