Abstract Background Surveillance imaging is used to detect local and/or distant recurrence following primary treatment of localised soft tissue sarcoma (STS), however evidence supporting optimal surveillance modality or frequency is lacking. We used prospectively collected sarcoma data to describe current surveillance imaging practice in patients with AJCC stage II and III extremity STS and evaluate its cost-effectiveness. Methods From three selected Australian sarcoma referral centres, we identified patients with stage II and III extremity STS treated between 2009 and 2013. Medical records were reviewed to ascertain surveillance imaging practices, including modality, frequency and patient outcomes. A discrete event simulation model was developed and calibrated using clinical data to estimate health service costs and quality adjusted life years (QALYs) associated with alternative surveillance strategies. Results Of 133 patients treated for stage II and III extremity STS, the majority were followed up with CT chest (86%), most commonly at 3-monthly intervals and 62% of patients had the primary site imaged with MRI at 6-monthly. There was limited use of chest-X-ray. A discrete event simulation model demonstrated that CT chest screening was the most cost effective surveillance strategy, gaining additional QALYs at a mean incremental cost of $30,743. MRI alone and PET-CT alone were not cost-effective, whilst a combined strategy of CT + MRI had an incremental cost per QALY gained of $96,556. Conclusions Wide variations were observed in surveillance imaging practices in this high-risk STS cohort. Modelling demonstrated the value of CT chest for distant recurrence surveillance over other forms of imaging in terms of cost and QALYs. Further work is required to evaluate cost-effectiveness in a prospective manner.