A theoretical model is outlined for predicting the time evolution of total mass, mean molecular weight, and drug release for the case of a spherical bulk-eroding microsphere, prepared by a double emulsification procedure and containing a hydrophilic drug, such as a protein or peptide. Explicit analytical formulae are derived for calculating the time evolution of measurable macroscopic characteristics, such as drug release or mean molecular weight. Microsphere hydration, polymer erosion, and drug release phases are each described. Results indicate that polymer degradation by only random-chain scission or only end scission (or unzipping) cannot explain experimentally observed kinetics of particle mass loss and molecular weight change; thus, a combined model (incorporating both random and end scission) is proposed. A general methodology for determining the microscopic transport coefficients (such as polymer degradation rate constant or drug diffusion coefficient) from erosion and release data is outlined. This paradigm is applied to the specific case of 50:50 poly(D,L-lactic-co-glycolic acid (PLGA) microspheres encapsulating glycoprotein 120 (gp 120), a candidate AIDS vaccine. Predictions permit comparisons with experimental data for mean weight- and number-averaged molecular weights, as well as for mass loss and protein release. Other comparisons are made with data appearing in the literature for release of tetanus toxoid from PLA and PLGA microspheres of variable molecular weight. Agreement between theory and experiment is observed.