Until recently, the concept of antibodies as in vivo therapeutics was still considered to be an exceedingly ambitious goal. However, in 2003, the situation has been completely transformed, with 14 FDA-approved monclonal antibodies (mAbs), 70 in late stage clinical (Phase II+) trials and > 1000 in preclinical development. The driving force behind this reversal in fortune has been advances in antibody engineering and the emergence of novel discovery techniques which overcame stability and immunogenicity issues that had blighted previous clinical trials of murine antibodies. For indications as diverse as inflammation, cancer and infectious disease, it is clear that unique properties of antibodies make them safe, effective and versatile therapeutics. These drugs can be used to neutralise pathogens, toxins and endogenous mediators of pathology. As cell targeting reagents, antibodies can be used to modulate cytoplasmic cascades or to 'tag' specific cells for complement- or effector-mediated lysis. Antibodies can also be modified to deliver toxic or modulatory payloads (small molecules, radionuclides and enzymes) and engineered to bind multiple epitopes (bispecifics) or even to have novel catalytic activity (abzymes). The modular structure of immunoglobulins and the availability of antibody fragment libraries also make it possible to produce variable-domain therapeutics (Fab, single-chain and domain antibodies). Although exhibiting less favourable kinetics in vivo, these fragments are simple to express and have an increased tissue penetration, making them especially useful as neutralising agents or in the delivery of payload. The number of approved antibodies is expected to increase arithmetically in the near term, as the platform is adopted as a valid alternative to small molecule discovery. This review provides an introduction to the antibody discovery process and discusses the past, present and future applications of therapeutic antibodies, with reference to several FDA-approved precedents.