Notes: Summary Insertin, a protein purified from chicken gizzard smooth muscle, has been shown to retard but not to inhibit actin polymerization at the barbed ends of actin filaments. This effect has been explained by a model in which insertin remains bound to the barbed ends of actin filaments and new actin molecules are inserted into filaments between the barbed ends and barbed end-bound insertin molecules. In this paper we discuss the mechanism of the insertion reaction on a molecular level. A number of simple models were devised and were judged by their agreement with available experimental data. In one class of models insertin was assumed to dissociate from filament ends and to re-associate with the ends. Actin monomers would then bind to a filament end between a dissociation and an association reaction of insertin. In one of the two proposed models in this class insertin binds to an ATP-containing terminal subunit with higher affinity than to an ADP-containing terminal subunit. Dissociation of insertin is brought about by ATP hydrolysis at the terminal filament subunit. Insertin was then thought to re-associate with a filament end following binding of an ATP-containing actin monomer to the filament end. In the other of the two models' insertin was assumed to occur in two conformations which bind to filament ends with different affinities. Association and dissociation of insertin is caused by interconversion between the two forms of insertin. Both models turned out to be incompatible with experimental data. All types of models in which retardation of actin polymerization is brought about by dissociation and re-association of insertin with filament ends can be excluded by a common argument. As 10 nM insertin retards polymerization of 2μm monomeric actin with maximal efficiency, the rate constant of binding of insertin to a filament end must be considerably higher (〉2μM/10 nM = 200-fold). As the rate of association of actin with a barbed end is almost diffusion-controlled, assembly of insertin with a filament end would have to exceed the rate of a diffusion-controlled reaction. In the other class of models it was assumed that insertin remains permanently bound to filament ends during association or dissociation of an actin molecule and to move towards the terminal subunit of filaments. These models are compatible with experimental data. Thus, models are favoured where insertin remains bound to filament ends during polymerization and depolymerization of actin.