Recent advances in combinatorial protein engineering have made it possible to develop non-Ig protein scaffolds that can potentially substitute for most whole antibody-associated properties. These protein scaffolds display most of the binding properties associated with the variable domain of antibodies. In theory, many different natural human protein backbones are suitable to be used as recombinant templates for engineering ; in practice however, only a few have yielded the necessary properties to be translated into << druggable biologicals >>. Amongst these properties, potential broad specificities towards any kind of target, ease of production, small size, good tolerability and low immunogenicity are essential. Intellectual property is another key issue. In this review, a particular emphasis will be given to the most validated non-Ig scaffolds that have reached the clinical development phase.