Antibiotics have been investigated as potential amnesic agents in different species using a variety of learning paradigms. The initial interest in antibiotics as amnesic agents stemmed from their effects on protein synthesis; they were viewed as an avenue for exploring various hypotheses concerning the role of protein synthesis in memory formation. The major classes of antibiotics used in investigations of this type exert effects on protein synthesis through different mechanisms and may have other biochemical effects as well. They also appear to have different amnesic effects. The glutarimides (cycloheximide and acetoxycycloheximide) are generally effective in producing retention deficits only when injected prior to initial training. Puromycin, on the other hand, appears to be consistently effective when injected intracerebrally 24 hours after training. The effects of actinomycin-D appear to be inconsistent. Close examination of the literature indicates that all of these antibiotics have "side effects" which should be an important caution to experimenters, but that sufficient control procedures have been used in a variety of experiments to justify an amnesic interpretation of antibiotic-induced deficits on retention tests. The data further suggest that antibiotic-induced amnesia is reversible by chemical or behavioral means, with the possible exception of the case in which puromycin is injected prior to learning. An intensive review of the studies relevant to the neurochemical effects of the major antibiotics suggests that protein synthesis per se is not the crucial correlate of their behavioral effects. There are probably different modes of chemical action corresponding to the different behavioral effects. Several possible mechanisms are discussed in detail, most notably glutarimide effects on neurotransmitter activity and the relationship of puromycin to brain peptides.