Objective. Non-primary radiation doses to normal tissues from proton therapy may be associated with an increased risk of secondary malignancies, particularly in long-term survivors. Thus, a systematic method to evaluate if the dose level of non-primary radiation meets the IEC standard requirements is needed. 
Approach. Different from the traditional photon radiation therapy system, proton therapy systems are composed of several subsystems in a thick bunker. These subsystems are all possible sources of non-primary radiation threatening the patient. As a case study, 7 sources in the P-Cure synchrotron-based proton therapy system are modeled in Monte Carlo (MC) code: tandem injector, injection, synchrotron ring, extraction, beam transport line, scanning nozzle and concrete reflection/scattering. To accurately evaluate the synchrotron beam loss and non-primary dose, a new model called the torus source model is developed. Its parametric equations define the position and direction of the off-orbit particle bombardment on the torus pipe shell in the Cartesian coordinate system. Non-primary doses are finally calculated by several FLUKA simulations.
Main results. The ratios of summarized non-primary doses from different sources to the planned dose of 2 Gy are all much smaller than the IEC requirements in both the 15 cm to 50 cm and 50 cm to 200 cm regions. Thus, the P-Cure synchrotron-based proton therapy system is clean and patient-friendly, and there is no need an inner shielding concrete between the accelerator and patient.
Significance. Non-primary radiation dose level is a very important indicator to evaluate the quality of a PT system. This manuscript provides a feasible Monte Carlo procedure for synchrotron-based proton therapy with new beam loss model. Which could help people figure out precisely whether this level complies with the IEC standard before the system put into clinical treatment. What’ more, the torus source model could be widely used for bending magnets in