The sublimation rate of frozen solutions was studied as a function of freezing rate, thickness of dried product (l), temperature, residual air pressure, and solute concentration. Data are presented for pure water, aqueous potassium chloride, aqueous povidone, and aqueous dobutamine hydrochloride-mannitol (System I). The resistance of the dried product to water vapor flow (Rp) was evaluated from the sublimation rate and the sample temperature. The primary experimental technique was based on freeze-drying a cylindrical microsample isothermally, with the sample suspended from one arm of a vacuum microbalance. Methodology to evaluate resistance data from vial freeze-drying experiments is also described. In separate experiments, samples in the form of a thin (15-microns) film were visually observed through a microscope during freeze-drying. Freeze-drying of most samples appeared to occur by water vapor escaping through open channels created by prior sublimation of ice. Contrary to the usual theoretical model, Rp is neither independent of temperature nor directly proportional to l. Rather, Rp decreases with increasing temperature and the l dependence is normally of the form Rp = (A0 + A1l)/(1 + A2l), where Ai (i = 0, 1, 2) are constants. In several cases, Rp is very large near l = 0, decreases sharply at l congruent to 0.1 cm, and obeys the above equation where l greater than 0.2 cm, a result suggesting an amorphous surface skin which cracks on desorption of water. The temperature dependence of Rp suggests that, as the sample temperature approaches the eutectic (or collapse) temperature, hydrodynamic surface flow of adsorbed water is an important flow mechanism.