In the absence of a definitive immunologic correlate of protection against shigellosis, promising Shigella vaccine candidates have been selected based on their ability to confer resistance against experimental challenge with wild-type Shigella in healthy adult volunteers. A limitation of this model has been the low and often variable attack rate of illness among controls, necessitating repeated inpatient studies to demonstrate statistically significant results. In this study, the Shigella challenge model was modified by using bicarbonate buffer instead of skimmed milk as the delivery vehicle to enhance survival of the ingested challenge inoculum. To determine the ability of the modified model to detect protective efficacy, 11 veteran volunteers (previously challenged with S. flexneri 2a in bicarbonate buffer) and 12 immunologically naive control subjects were challenged with 1.4 x 10(3) c.f.u. S. flexneri 2a. Shigellosis occurred in 3 veterans and 11 control subjects (27 vs 92%, p = 0.003), yielding a protective efficacy of 70%. Dose response was evaluated in an additional seven naive subjects who were inoculated with a log lower (1.4 x 10(2) c.f.u.) S. flexneri 2a and had a significantly diminished attack rate of shigellosis (317 (43%) vs 11/12 (92%), p = 0.04). These findings indicate that the modified bicarbonate challenge model using an inoculum of 10(3) c.f.u. is a safe, repeatable, and valid method of selecting Shigella vaccines and other immunoprophylactic agents that are likely to confer protection against natural shigellosis.