Biosensor sensitivity and selectivity essentially depend on the properties of the biorecognition elements to be used for analyte binding. Immunosensors employing antibodies as binding protefìts are effective tools for the analysis of a wide variety of analytes from the group of antigens and haptens. Genetic engineering provides an elegant way not only for the generation of virtually unlimited amounts of biorecognition molecules, but also for the alteration of existing properties and supplementation with additional functions. Different strategies for the synthesis of antibody fragment libraries, followed by the selection of specific antibody variants, were examined An antibody library was derived from a set of B cells. Chain shuffling of the antibody heavy and light chains provided the improved binders. An ELISA was achieved for the herbicide atrazine with an IC50 of 0.9 µg/1 and a detection limit of 0.2 µg/1, which corresponds to an equilibrium dissociation constant of 7.46×10−10 M as determined by a surface plasmon resonance-based biosensor. The close relations between the optimization of recombinant antibodies by evolutionary strategies and genetic algorithms are considered.