When trying to obtain a cDNA clone to a novel protein, the only handle one may have is an antibody that recognizes the protein of interest. Consequently, the obvious approach is to screen a λ-phage expression library using the antibody. In general, polyclonal sera give better results, but a mixture of monoclonal antisera could be used instead. Fortunately, a number of convenient commercial vectors are available that can be used to generate an expression library where the cloned inserts are induced to produce protein within bacteria, usually as a fusion protein with β-galactosidase. Alternatively, it may be possible to buy a library of a cell line or tissue known to express the protein of interest in reasonable abundance. The library should have a complexity of at least 10⁶, since in theory only one in six inserts will clone in the orientation and frame required to produce protein, although modern vectors often allow directional cloning. Thus, one in three of the clones to the protein of interest may be recognized by the antibody. Attention should also be given to the average size of the inserts, particularly for a large protein, since even polyclonal sera to the full protein may only recognize one epitope, which could lie at the N-terminus. Random-primed libraries tend to contain more N-terminal sequences and fewer extraneous sequences such as long 3′-untranslated regions than oligo-dT-primed libraries, but the former are rarely directional. A suitable compromise may be to screen a mixed library that has been generated using both priming methods, and many of these are available commercially.