Phosphorylation of the high mannose-type oligosaccharides attached to newly synthesized acid hydrolases occurs in two sequential steps within the endoplasmic reticulum and the Golgi apparatus, and the products generated at the two sites differ with respect to the location of the phosphorylated mannose residue. To investigate the mechanism of this two-step phosphorylation, biosynthesis of the Man-6-P recognition marker was studied in class E Thy-1- and J774 cells metabolically labeled with [2-3H]mannose. Class E Thy-1- cells produce truncated high mannose oligosaccharides that lack 4 mannose residues from the alpha 1,6-branch of the core beta-linked mannose residue; three of the missing residues are potential phosphorylation sites. Acid hydrolases produced by these mutant cells were phosphorylated on the alpha 1,3-branch of the truncated oligosaccharide even when transport to the Golgi apparatus was inhibited. J774 cells produce normal high mannose oligosaccharides, but they secrete a large percentage of their newly synthesized acid hydrolases. The secreted enzymes contained primarily diphosphorylated units in which a phosphate was positioned to both the alpha 1,3- and alpha 1,6-branches of the core beta-linked mannose. J774 cells treated with deoxymannojirimycin continued to phosphorylate and to secrete acid hydrolases. The secreted hydrolases, however, contained only monophosphorylated oligosaccharides in which the phosphate was restricted to the alpha 1,6-branch. These results indicate that mannose residues within high mannose oligosaccharides impose constraints on the phosphorylation of their composite structures. We conclude that the two-step phosphorylation occurs as a result of a common phosphotransferase at both the pre-Golgi and Golgi locations and a change in the conformation of the oligosaccharides attached to the acid hydrolases through the action of Golgi-associated alpha-mannosidase I.