Transforming growth factor (TGF)-beta signalling plays a dichotomous role in tumour progression, acting as a tumour suppressor early and as a pro-metastatic pathway in late-stages. There is accumulating evidence that advanced-stage tumours produce excessive levels of TGF-beta, which acts to promote tumour growth, invasion and colonisation of secondary organs. In light of the pro-metastasis function, many strategies are currently being explored to antagonise the TGF-beta pathway as a treatment for metastatic cancers. Strategies such as using large molecule ligand traps, reducing the translational efficiency of TGF-beta ligands using antisense technology, and antagonising TGF-beta receptor I/II kinase function using small molecule inhibitors are the most prominent methods being explored today. Administration of anti-TGF-beta therapies alone, or in combination with immunosuppressive or cytotoxic therapies, has yielded promising results in the preclinical and clinical settings. Despite these successes, the temporal- and context-dependent roles of TGF-beta signalling in cancer has made it challenging to define patient subgroups that are most likely to respond, and the therapeutic regimens that will be most effective in the clinic. Novel mouse models and diagnostic tools are being developed today to circumvent these issues, which may potentially expedite anti-TGF-beta drug development and clinical application.