Tablets (300mg) having two different surface areas were prepared from spray-dried particles of curdlan (100mg)/theophylline (200mg). Drug release from the tablets was studied in vitro and in vivo. The in vitro drug release from a tablet with a larger surface area (Tab L) was faster than that with a smaller one (Tab S). The water uptake of Tab L was larger than that of Tab S, probably due to the difference in the tablets' surface areas. However, the water uptake was not a rate-determining step for the drug release from curdlan tablets containing a large amount of theophylline. A straight line was obtained when release % was plotted vs. time. The slope of each curve was calculated as 0.59 for Tab L and 0.58 for Tab S. This indicates that the release mechanism is non-Fickian diffusion controlled. In addition, the curdlan tablets or theophylline powder were administered orally to 5 healthy volunteers, and saliva concentrations of theophylline were determined. Each saliva concentration was converted to plasma concentration using the saliva to plasma ratio of the drug in each subject. The AUC of Tab L was nearly the same as that of powder, while the AUC of Tab S was smaller than that of powder. The mean residence times (MRTs) of theophylline powder, Tab S and Tab L were 11.1±1.5, 25.4±6.3 and 17.1±1.5h (N=4-5, mean±S.D.), respectively. The mean dissolution times (MDTs) of Tab L in vivo and Tab S in vivo were 5.0±2.1 (N=5, mean±S.D.) and 13.9±5.4h (N=4, mean±S.D.), respectively. On the other hand, the MDTs of Tab L in vitro and Tab S in vitro were 4.8 and 11.2h, respectively. In vivo drug release was very similar to in vitro drug release in both tablets. The lower bioavailability of Tab S suggested that the drug release had not been completed during the gastrointestinal transit period. Tab L would thus be a better controlled release form than Tab S.