In vertebrate smooth muscles, phosphorylation of the regulatory light chain appears to be necessary for actin activation of the Mg-ATPase activity and for the in vitro assembly of myosin into filaments. From a correlation between the degree of phosphorylation and enzymatic activity, it was suggested that both myosin heads must be phosphorylated before either head could be activated by actin, and that phosphorylation of filamentous myosin occurred in a negatively cooperative manner (Persechini, A., and Hartshorne, D. J. (1981) Science 213, 1383-1385; Ikebe, M., Ogihara, S., and Tonomura, Y. (1982) J. Biochem. (Tokyo) 91, 1809-1812; Sellers, J. R., Chock, P. B., and Adelstein, R. S. (1983) J. Biol. Chem. 258, 14181-14188). Here we have determined the mechanism of phosphorylation by separating dephosphorylated and phosphorylated myosin species based on their different structural properties in the minifilament buffer system (5 mM citrate, 22 mM Tris). Fully phosphorylated myosin remained assembled as minifilaments in 1 mM Mg-ATP, but dephosphorylated myosin dissociated to a mixture of folded monomers and dimers. Gel filtration was used to separate these two structures. At intermediate levels of phosphorylation, the relative amount of myosin that formed minifilament and dimer and the degree of phosphorylation of the separated species relative to the initial level of phosphorylation was measured. From these data, it was possible to deduce that singly and doubly phosphorylated myosin remained assembled in the presence of nucleotide. Myosin molecules with 0, 1, or 2 heads phosphorylated could also be separated by nondenaturing gel electrophoresis. The amount of myosin which formed each species was quantitated as a function of phosphorylation. Results from the combined approaches are consistent with a model in which light chain kinase randomly phosphorylates myosin, independent of the state of aggregation of the myosin.