The early events in the thrombin-induced formation of fibrin have been studied by the use of stopped-flow multiangle laser light scattering (SF-MALLS). This technological advancement has allowed the recovering, as a function of time with a resolution of about 0.5 sec, of the mean square radius of gyration (R-g(2))(z) and of the molecular weight M-w, and to place an upper bound to the values of the mass/unit length M-L. The ionic strength, pH and salt type conditions investigated were all close to physiological, starting with a 50 mM Tris, 104 mM NaCl, pH 7.4 buffer (TBS), to which either 1 mM EDTA-Na-2 or 2.5 mM CaCl2 were also added. Fibrinogen was 0.2-0.3 mg/ml and rate-limiting concentrations of thrombin were used (0.05-0.25 NIH units/mg fibrinogen). By plotting (R-g(2))(z) and M-L versus M-w on log-log scales, runs proceeding at different velocities and under different solvent conditions could be compared and confronted with model curves. It was found that: (1) within this thrombin range, the mechanism of association does not depend on its concentration, nor on the buffers employed; (2) the (R-g(2))(z) versus M-w curves could all be reasonably fitted with a bifunctional polycondensation scheme involving semiflexible worm-like, double-stranded, half-staggered polymers with persistence length between 200-600 nm, provided that a ratio Q = 16 between the rate of release of the two fibrinopeptides A was employed; (3) the M-L versus M-w data seemed more compatible with lower Q values (4 < Q < 8), but their uncertainty prevented a better assessment of this issue; the formation of fibrinogen-fibrin monomer complexes may also play a role in the polymer distributions; (4) in the very early stages (e.g., when M-w < 7 x 10(5)), the (R-g(2))(z) versus M-w data were fitted well only in TBS and at the lowest thrombin concentration, suggesting that a transient, either sequential or concurrent fast second mechanism, involving longer and thinner polymers, may be at work.