Animal models are important in predicting the efficacy in humans of antimicrobial agents for various disease conditions, including endocarditis and meningitis. Screening models are useful in assessing antibiotic effectiveness and toxicity; their advantages include simplicity, a reproducible course of infection, a well-defined therapeutic end point, and low cost. However, the inoculum size, the virulence of the organism, and the production of beta-lactamases can have important effects on outcome and must be considered in the interpretation of data obtained from such models. Discriminative models are those designed to mimic human disease as closely as possible with respect to infectious inoculum, host response, and course of disease. Each drug's pharmacokinetics must be carefully documented before being extrapolated to humans. Rigid criteria must be established to minimize misinterpretation of results from animal studies before conclusions from in vivo animal models are applied to human disease.