Background: Despite being the sixth most common cause of cancer death in Australia, cancer of unknown primary (CUP) site remains poorly understood. Aims: To describe practices relating to the diagnosis, investigation, classification, communication and management of CUP among medical oncologists. Methods: We invited all members of the Medical Oncology Group of Australia to participate in a national, anonymous online survey about CUP. The survey collected data regarding diagnosis acceptance, diagnostic tests, treatment protocols and communication practices around the diagnosis of CUP. Results: Three hundred and two oncologists were invited and 86 (28%) completed the survey. Eighty (93%) respondents were directly involved in the assessment of patients with CUP. Eighty-five (99%) respondents were prepared to make a diagnosis of CUP if, after appropriate diagnostic tests, the primary location could not be ascertained. Eighty-three percent would assign a primary site to obtain Pharmaceutical Benefits Schedule funding of medical therapy. Sixty-two percent did not have a specific treatment protocol designed for CUP. The majority of oncologists used serum tumour markers and computed tomography scans in the initial work-up, while 43% indicated they would use a positron emission tomography scan in the majority of cases. The majority would arrange mammography in female patients. Thematic analysis of responses to open-ended questions about how CUP is described identified little consistency in the language being used. Conclusion: The approach to diagnosis, investigation and management of CUP by medical oncologists in Australia is variable. Many preferred to estimate the primary site and treat accordingly. Pharmaceutical Benefits Schedule restrictions may encourage the practice of ‘best guessing’.