Most cancer‑related deaths are due to metastatic disease, primarily reflecting the paucity of effective treatments once cancer has spread beyond the primary organ. Investigations into the molecular and cellular processes underlying the metastatic cascade are vital for streamlining therapeutic approaches directed against disseminated disease. To this end, extensive efforts have been devoted to the development of innovative and clinically relevant in vivo preclinical models that recapitulate metastasis in patients. The major platforms currently utilized include chemical induction, transplantation and genetically engineered mouse models. These models are used primarily to: (1) investigate the mechanisms regulating metastasis, (2) identify and validate novel prognostic and therapeutic targets and (3) test novel therapies. Here, we delineate the advantages and disadvantages of in vivo metastasis platforms currently in use, to aid investigators with the selection of most relevant models for their research examining specific aspects of metastasis and/or the efficacy of novel therapies. We briefly discuss the limited ability of current models to accurately predict drug response in the clinic and the restricted range of molecular tumor subtypes that can be modeled. These remain important issues and represent key challenges to improve the clinical relevance of the next generation of metastasis models.