Abstract. Direct electron microscopic examination confirms that short actin filaments rapidly anneal endto-end in vitro, leading over time to an increase in filament length at steady state. During annealing of mixtures of native unlabeled filaments and glutaraldehyde-fixed filaments labeled with myosin subfragment-1, the structural polarity within heteropolymers is conserved absolutely. Annealing does not appear to require either ATP hydrolysis or the presence of exogenous actin monomers, suggesting that joining occurs through the direct association of filament ends. During recovery from sonication the initial rate of annealing is consistent with a second-order reaction involving the collision of two filament ends with an apparent annealing rate constant of 107 M-Is -I. This rapid phase lasts <10 s and is followed by a slow phase lasting minutes to hours. Annealing is calculated to contribute minimally to filament elongation during the initial stages of self-assembly. However, the rapid rate of annealing of sonicated fixed filaments observed in vitro suggests that it may be an efficient mechanism for repairing breaks in filaments and that annealing together with polymer-severing mechanisms may contribute significantly to the dynamics and function of actin filaments in vivo.steady state in vitro, preparations of actin filaments (Nakaoka and Kasai, 1969;Kawamura and Maruyama, 1970) and microtubules (Mitchison and Kirschner, 1984) grow longer and become progressively fewer in number. While monomer exchange at polymer ends is clearly an important factor in determining the growth and stability of polymers of actin and tubulin, it is likely that polymer fragmentation and annealing are also involved in determining the lengths and numbers of polymers at steady state. In recent studies of microtubule assembly, we demonstrated that microtubule polymers can rapidly join end-to-end (anneal) (Rothwell et al., 1986) and that, under certain conditions, mechanisms of polymer annealing and monomer exchange both play important roles in determining microtubule dynamics in vitro (Rothwell et al., 1987). In the present paper, we examine the role of annealing in the assembly of actin filaments.There are divided opinions regarding the existence of both fragmentation and annealing during actin filament assembly. There is irrefutable evidence that actin filaments can break into smaller pieces when subjected to strong mechanical forces (Asakura, 1961), although it is less certain that thermal energy alone is strong enough to break filaments in samples that have not been stirred. Mathematical models for the spontaneous polymerization of actin fit the time course of assembly better when a term for spontaneous fragmentation is included (Wegner, 1982;Wegner and Savko, 1982;Cooper et al., 1983). However, the contribution of fragmentation and annealing to filament assembly are thought to be complex (Frieden and Goddette, 1983). In these papers, a fragmentation term was not required for all polymerization conditions and there were contradiction...