Microspheres of fixed poly(D,L-lactic acid) (PDLLA) composition--Resomer R104:R202H (30:70)--containing 20% w/w rifampicin have been spray-dried from a range of acetonic, halogenated, and solvent mixtures thereof under constant process conditions to examine the influence of solvent selection on microsphere characteristics. Solubility of the polymer composite in the studied solvents determined the kinetics of polymer deposition during drying. Viscosity studies provided an indirect index of solvent power in ascending order: acetone (ACT) < dichloromethane (DCM) < chloroform (CFM) < halothane (HAL). Accordingly, poorer acetonic solvents produced a more open, porous matrix of increased mean diameter, whereas DCM, CFM and HAL generated more coherent matrices of greater density and elevated glass transition temperature, which significantly retarded drug release. Yield generally increased in parallel with solvent strength and microsphere density consistent with the proposed generalized particle formation mechanism. Residual solvent also increased with particle density, both parameters being interrelated and dictated by the inherent affinity of the polymer composite for individual solvents. In turn, the position of glass transition temperature (Tg) and the quantity of associated polymer stress-relaxation were a direct function of amount and persistence of organic residue. The magnitude of these changes determined the relative rates and extents of microsphere ageing, as measured by drug release studies. In general, rate of drug release increased with Tg, after corrections were made for specific surface area (r2 = 0.963). Overall, solvent choice for spray-drying has a remarkable influence on microsphere characteristics and, accordingly, technological as well as toxicological considerations should be paid during selection of same.