For more than 50 years, researchers have tested a variety of killed, attenuated, and subunit preparations for control offish diseases. The earliest fish vaccines used killed preparations containing whole bacteria, viruses, or parasites and today, several bacterins have become commercially successful with more expected as improved delivery systems and adjuvants are realized. Live, attenuated vaccines have been developed by serial passage of a pathogen in culture or by using naturally occurring mutants and cross-reacting strains. These generally offer excellent protection and are cost-effective, but concerns about residual virulence or their effects on other aquatic species make them difficult candidates for licensing. In recent years, the tools of molecular biology have been applied to construction of a variety of recombinant, engineered, or subunit vaccines for fish. Among the approaches to be discussed are: attenuated strains of viruses and bacteria created by deletion of specific genes associated with virulence, in vitro synthesis of protective antigens from genes cloned into E. coli or baculovirus expression systems, chemical synthesis of peptides that represent protective epitopes, and direct immunization with DNA coding for protective antigens. Preparations representing each of these approaches have been tested in laboratory or field trials with various results and such vaccines promise to be safe and relatively inexpensive if they are able to provide protection when delivered by immersion. A significant advantage of genetically engineered vaccines is the ability to construct multivalent preparations that can protect fish against several pathogens or different strains of the same pathogen. While many of these novel vaccine strategies have been effective at stimulating specific immunity in the laboratory, more work is needed to develop better delivery systems and to overcome potential regulatory concerns.