Particularly for proteins that may be damaged by overdrying, the distribution of residual moisture in the dried product may be as important as the mean water content. Protein at the top of the cake could be "overdried" even though the mean water content was optimal. Theoretical studies suggest that, particularly for large fill depths, the residual moisture near the top of the cake is much lower than the moisture near the bottom, at least during and immediately after freeze drying. This work is an experimental study of the distribution of moisture within a vial as a function of time in secondary drying (mean residual water content of about 3%-6%) for dextran, human serum albumin, and bovine somatotropin. A core sample of the dried product was taken and sectioned into "top," "middle," and "bottom" sections. These three "core sections" and the remaining "outer" section (i.e., near the vial wall) were assayed for moisture. In general, the moisture content in the top section is less than the moisture content in the bottom section, but the "outer" section is consistently found to be lowest in moisture content. This observation suggests that drying was faster along the walls of the vial than in the core region, resulting in a highly curved ice-vapor interface. It is proposed that faster drying along the vial walls is a result of the observed product shrinkage during drying which provides a low resistance pathway for vapor escape along the vial wall. Product temperature measurements support the above speculation.