Performance in H-mode DT plasmas in ITER with various choices of heating systems are predicted and compared. Combinations of external heating by negative ion neutral beam injection (NNBI), ion cyclotron range of frequencies and electron cyclotron heating are assumed. Scans with a range of physics assumptions about boundary temperatures in the edge pedestal, alpha ash transport and toroidal momentum transport are used to indicate effects of uncertainties. Time-dependent integrated modelling with the PTRANSP code is used to predict profiles of heating, beam torque and plasma profiles. The GLF23 model is used to predict temperature profiles. Either GLF23 or the assumption of a constant ratio for χϕ/χi is used to predict toroidal rotation profiles driven by the beam torques. Large differences for the core temperatures are predicted with different mixes of the external heating during the density and current ramp-up phase, but the profiles are similar during the flat-top phase. With χϕ/χi = 0.5, the predicted toroidal rotation is relatively slow and the flow shear implied by the pressure, toroidal rotation and neoclassical poloidal rotation are not sufficient to cause significant changes in the energy transport or steady state temperature profiles. The GLF23-predicted toroidal rotation is faster by a factor of six, and significant flow shear effects are predicted. Heating mixes with more NNBI power are predicted to have up to 20% higher fusion power during steady state phases. This advantage is decisive in some cases where the physics assumptions are close to marginal or critical values. L-mode plasmas are predicted having QDT ≃ 2–4.