Studies of the dependence of actin polymerization on thermodynamic parameters are important for understanding processes in living systems, where actin polymerization and depolymerization are crucial to cell structure and movement. We report measurements of the extent of polymerization, Phi, of rabbit muscle actin as a function of temperature [T=(0-35) degreesC], initial G-actin concentration [[G(0)]=(1-3) mg/ml], and initiating salt concentration [[KCl]=(5-15) mmol/l with bound Ca2+], in H2O and D2O buffers and in the presence of adenosine triphosphate (ATP). A preliminary account of the data and analysis for H2O buffers has appeared previously [P. S. Niranjan, J. G. Forbes, S. C. Greer, J. Dudowicz, K. F. Freed, and J. F. Douglas, J. Chem. Phys. 114, 10573 (2001)]. We describe the details of the studies for H2O buffers, together with new data and analysis for D2O buffers. The measurements show a maximum in Phi(T) for H2O buffers and D2O buffers. For H2O buffers, T-p decreases as either [G(0)] or [KCl] increases. For D2O buffers, T-p decreases as [KCl] increases, but T-p is not monotonic in [G(0)]. The measurements are interpreted in terms of a Flory-Huggins-type lattice model that includes the essential steps: monomer activation, dimerization of activated species, and propagation of trimers to higher order polymers. The competition between monomer activation and chain propagation leads to the observed nonmonotonic variation of Phi(T). The actin polymerization in D2O buffer differs considerably from that in the H2O buffer and underscores the significant deuterium effect on hydrophobic interactions and hydrogen bonding in the polymerization process. (C) 2003 American Institute of Physics.