Targeted therapies, especially monoclonal antibodies, have reached an increasing importance in oncology. High-throughput techniques have allowed the identification of numerous transcripts, proteins, or non-protein antigens, which have generated the concept of immunome. This epitope library constitutes a huge reservoir of candidate antigens susceptible to become some say the target of an antibody for passive immunotherapy. However, the conception and development of a therapeutic antibody represent a very important investment, both in terms of human power and finance, such that there is a requirement for an early identification of the best candidates among the potential target antigens. Among multiple criteria, the function of the antigen is crucial when it has been identified. A receptor antigen can be targeted by an agonistic or an antagonistic antibody, according to what is sought. When the antigen function is unknown, a therapeutic antibody can be useful, for instance through induction of apoptosis or through accrual of immuno-competent cells, via its Fc portion (complement-dependent cytotoxicity or antibody-dependent cytotoxicity). Other antibody features, unrelated to its function, can also be exploited, such as its internalisation or its translocation in membrane lipid rafts. The expression of the target antigen may also be crucial, in terms of localisation and levels, as is its tumour specificity, which can influence the efficacy and toxicity of the targeting antibody. The multiplicity of the factors to be taken into account and the complexity of the mechanism of action of therapeutic antibodies renders the choice of a target antigen a hazardous bet. Very often, this is only when the clinical efficacy of a targeting antibody is demonstrated that the antigen can be considered as a good target.