In this paper, the synthesis problem of complex nonisothermal reactor networks is addressed. The focus is on developing a superstructure of alternatives for the nonisothermal reactor system, and subsequently formulate and solve the synthesis problem according to the proposed general structure. The nonisothermal case features alternatives related to the type of temperature control for the reactors and includes options for directly or indirectly intercooled or interheated reactors. The approach is applicable to any homogeneous exothermic or endothermic complex reaction and the synthesis problem, formulated as a mixed integer nonlinear programming (MINLP) problem, can handle both thermodynamic and economic objective functions. The solution of the resulting MINLP problem provides information about the optimal type of temperature control, the optimal temperature profile, the feeding, recycling and bypassing strategy as well as the optimal type and size of the reactor units (CSTR, PFR) to be used. The efficiency of the method is illustrated with three different examples considering complex nonisothermal reaction mechanisms.