Objective: To evaluate the effectiveness of exercise for improving bone-related outcomes among cancer survivors. Design: Systematic review and meta-analysis. Methods: An electronic search using the following databases: SPORTDiscus, Science Direct, CINAHL, MEDLINE, Cochrane, Pubmed, Ebscohost, ProQuest Nursing and Allied Health Source. Randomised, controlled, exercise trials involving cancer survivors were eligible. Effect data on bone mineral content (BMC) and density (BMD) outcomes were extracted. Risk of bias was assessed using the Physiotherapy Evidence Database tool. Standardised mean differences (SMD) were calculated to compare differences between exercise and usual care. Subgroup analyses were conducted to assess whether effect differed by exercise mode, intervention length, supervision, treatment, cancer type and risk of bias. Results: Twenty-six trials were included, with intervention durations ranging between 12 weeks and 2 years. Most trials involved breast cancer (n = 13, 50%), and most interventions were supervised (n = 18, 69%) and evaluated mixed-mode (i.e., combined aerobic and resistance) exercise (n = 13, 50%). Significant effects in favour of exercise (aerobic, resistance, mixed-mode and other exercise) were observed for whole body BMD, hip BMD, trochanter BMD and femoral neck BMD (SMD range: 0.19–0.39, all p < 0.05) compared to usual care. Conclusion: Participation in various modes (aerobic, resistance, mixed-mode and other) of supervised and unsupervised exercise is associated with improvements in BMD. The present results provide evidence for clinicians and other health care professionals (e.g., exercise physiologists and physiotherapists) to recommend exercise for cancer survivors to prevent bone loss during and following treatment.