This work experimentally and theoretically investigates the use of bifunctional initiators in the synthesis of high-impact polystyrene (HIPS). The experimental design involved a series of nonisothermal bulk polymerizations of styrene (St) in the presence of polybutadiene (PB). The performance of three commercial initiators [2,5-dimethyl-2,5 bis(2-ethylhexanoyl peroxy) hexane or L-256; 2,5 dimethyl-2,5 bis(benzoyl peroxy) hexane or L-118; and ethyl 3,3 di(t-butyl peroxide) butirate or L-233] were compared to the performance of a standard monofunctional initiator (terbutylperoctoate or TPBO), and to the blank case (i.e., without initiator). From samples taken along the prepolymerization period, the phase inversion point and the 30% conversion point were estimated. For the final product, the free polystyrene (PS) molecular weights and the St grafting efficiency were measured. A mathematical model was developed that predicts the evolution of the MWDs for the free PS, the residual PB, and the graft copolymer, together with the chemical composition distribution for the total graft copolymer. Compared to the monofunctional case, the L-256 initiator induces phase inversion and rubber grafting at low conversions. Also, it shortens the prepolymerization times by around 38%, without affecting the molecular characteristics of the final product. L-118 also shortens prepolymerization time with respect to TBPO; but is not as effective as L-256 or TBPO in promoting rubber grafting. At the polymerization end, the final molecular characteristics are practically independent of the initiator type because most of the polymerization is induced by monomer initiation. Due to its slow decomposition rate, the L-233 initiator is less effective that TBPO for reducing prepolymerization times and for promoting phase inversion.