Abstract Context: Two million people across the UK are living with cancer, often experienced as a long-term condition. They may have unmet needs after active treatment. Rehabilitation aims to address these needs, maximise psychological and physical function, and enable minimum dependency regardless of life expectancy. Objectives: We aimed to test in a randomised controlled trial the clinical and cost effectiveness of a rehabilitation intervention for patients with advanced recurrent cancer. Methods: We conducted a two-arm wait-list-control randomised trial of a complex rehabilitation intervention delivered by a hospice-based multi-disciplinary team versus usual care for active, progressive, recurrent haematological and breast malignancies with follow-up at 3 months. The primary outcome was the psychological subscale of the Supportive Care Needs Survey (SCNS). Secondary outcomes were other domains on SCNS, psychological status, continuity of care, quality of life and resource use. Results: Forty-one participants were enrolled and 36 completed the trial. The primary outcome was significantly lower in the intervention arm (adjusted difference -16.8, 95%CI -28.4 to -5.3, p=0.006). The SCNS physical and patient care subscales (-14.2, 95%CI -26.2 to -2.2, p=0.02, and -7.4, 95%CI -13.7 to -1.1, p=0.02, respectively), and self-reported health state (12.8, 95%CI 3.2 to 22.4, p=0.01) also differed significantly. The incremental cost-effectiveness ratio was £19,390 per QALY. Conclusions: This intervention significantly reduced unmet needs for cancer survivors and it is likely that it is cost-effective. Despite small numbers the main effect size was robust. We recommend implementation alongside evaluation in wider clinical settings and patient populations.