We have previously reported that self-diffusion of inert tracer particles in the cytoplasm of living Swiss 3T3 cells is hindered in a size-dependent manner (Luby-Phelps, K., D.L. Taylor, and F. Lanni. 1986. J. Cell Biol. 102:2015-2022; Luby-Phelps, K., P.E. Castle, D.L. Taylor, and F. Lanni. 1987. Proc Natl. Acad. Sci. USA. 84:4910-4913). Lacking a theory that completely explains our data, we are attempting to understand the molecular architecture responsible for this phenomenon by studying tracer diffusion in simple, reconstituted model systems. This report contains our findings on tracer diffusion in concentrated solutions of Ficoll 70 or Ficoll 400, in solutions of entangled F-actin filaments, and in solutions of entangled F-actin containing a background of concentrated Ficoll particles or concentrated bovine serum albumin (BSA). A series of size-fractionated fluorescein-Ficolls were used as tracer particles. By fluorescence recovery after photobleaching (FRAP), we obtained the mean diffusion coefficients in a dilute, aqueous reference phase (Do), the mean diffusion coefficients in the model matrices (D), and the mean hydrodynamic radii (RH) for selected tracer fractions. For each model matrix, the results were compared with similar data obtained from living cells. As in concentrated solutions of globular proteins (Luby-Phelps et al., 1987), D/Do was not significantly size-dependent in concentrated solutions of Ficoll 400 or Ficoll 70. In contrast, D/Do decreased monotonically with increasing RH in solutions of F-actin ranging in concentration from 1 to 12 mg/ml. This size dependence was most pronounced at higher F-actin concentrations. However, the shape of the curve and the extrapolated value of D/Do in the limit, RH----O did not closely resemble the cellular data for tracers in the same size range (3 less than RH less than 30 nm). In mixtures of F-actin and Ficoll or F-actin and BSA, D/Do was well approximated by D/Do for the same concentration of F-actin alone multiplied by D/Do for the same concentrations of Ficoll or BSA alone. Based on these results, it is possible to model the submicroscopic architecture of cytoplasm in living cells as a densely entangled filament network (perhaps made up of F-actin and other filamentous structures) interpenetrated by a fluid phase crowded with globular macromolecules, which in cytoplasm would be primarily proteins.