Selectively-infective phage (SIP) is a novel methodology for the in vivo selection of interacting protein-ligand pairs. It consists of two components, (1) a phage particle made non-infective by replacing its N-terminal domains of geneIII protein (gIIIp) with a ligand-binding protein, and (2) an “adapter” molecule in which the ligand is linked to those N-terminal domains of gIIIp which are missing from the phage particle. Infectivity is restored when the displayed protein binds to the ligand and thereby attaches the missing N-terminal domains of gIIIp to the phage particle. Phage propagation is thus strictly dependent on the protein-ligand interaction. We have shown that the insertion of β-lactamase into different positions of gIIIp, mimicking the insertion of a protein-ligand pair, led to highly infective phage particles. Any phages lacking the first N-terminal domain were not infective at all. In contrast, those lacking only the second N-terminal domain showed low infectivity irrespective of the presence or absence of the F-pilus on the recipient cell, which could be enhanced by addition of calcium. An anti-fluorescein scFv antibody and its antigen fluorescein were examined as a protein-ligand model system for SIP experiments. Adapter molecules, synthesized by chemical coupling of fluorescein to the purified N-terminal domains, were mixed with non-infective anti-fluorescein scFv-displaying phages. Infection events were strictly dependent on fluorescein being coupled to the N-terminal domains and showed a strong dependence on the adapter concentration. Up to 106 antigen-specific events could be obtained from 1010 input phages, compared to only one antigen-independent event. Since no separation of binders and non-binders is necessary, SIP is promising as a rapid procedure to select for high affinity interactions.