The insertion of newly synthesized protein molecules into the membrane of the secretory granule of the rat parotid gland was studied by in vivo labeling with [3-H]-proline and [3-H]leucine. 2 h after the injection of the amino acid into the rat, the membrane fraction isolated from the secretory granules was found to be highly labeled with proline but only slightly labeled with leucine. The ratio of proline label in the granule membrane to that in the granule's secretory content was roughly equivalent to the ratio of total proline in the proteins of these two fractions. In contrast the ratio of leucine label in the membrane to that in the secretory content was much less than would be expected from the relative amount of leucine in both fractions. Separation of the proteins of the granule membrane by gel electrophoresis in presence of sodium dodecylsulfate showed that a considerable amount of these proteins was unlabeled. The labeled proteins could be selectively extracted from the membrane by 0.15 M Nacl solution or by dilute buffer at pH 4.5. These extracted proteins were found to contain a high proportion of proline residues and a negligible amount of leucine residues. In the extract proline constituted 36 mole % of the total amino acids. Proline plus glycine plus glutamic acid constituted more than 80 mole % and leucine constituted about 1 mole% of the total amino acids. Further analyses by gel electrophoresis in presence of sodium dodecylsulfate showed that the fractions of secretory granule membrane and secretory granule content are relatively free of contamination by proteins from other subcellular structures. It is suggested that the proteins which will constitute the mature secretory granule are transported to the site of final assembly by two pathways. The proline-rich proteins are transported to the site of assembly in close coordination with all the exportable proteins. The other membrane proteins arrive by a different pathway. Two alternative mechanisms are suggested to explain the finding that a considerable part of the membrane proteins are not labeled. I. The pathway of the intracellular transport of the unlabeled membrane proteins is similar to that of the secretory proteins but the newly synthesized membrane protein molecules are diluted in a large intermediate pool--the GOLgi complex. II. The proteins that did not get labeled are derived by a process of reutilization, from membranes of granules which have previously discharged their content in the process of secretion.