In the absence of antimicrobial therapy, bacteria such as Bacteroides fragilis, Escherichia coli and proteus mirabilis may persist within an intra-abdominal abscess in the presence of large numbers of neutrophils which, under optimal conditions in vitro, can readily phagocytose and kill the same bacterial strains. Neutrophils taken from abscesses induced by gram-negative bacteria such as those above contain viable organisms. On incubation in vitro in the presence of serum, these neutrophils kill the bacteria phagocytosed in the abscess poorly, if at all, yet can readily kill organisms added in vitro. To determine possible mechanisms that might explain this, we examined the bactericidal activity in vitro of neutrophils from a range of abscesses induced by one or two species of bacteria plus an abscess-potentiating agent, bran. The organisms studied were B. fragilis, E. coli, P. mirabilis and Staphylococcus aureus. The killing in vitro of E. coli and P. mirabilis, engulfed within an abscess, was significantly less than that of the same organisms when they were added to the in-vitro assay. In contrast, the killing of S. aureus was similar, whether engulfed in vivo or in vitro. However, S. aureus was less susceptible to phagocytosis and killing in vitro than P. mirabilis or E. coli, and the killing of S. aureus during in-vitro incubation of neutrophils that had engulfed the organism within the abscess was similar to that of the gram-negative bacteria engulfed within the abscess. We examined whether organisms phagocytosed in vivo were within neutrophils which had diminished or no expression of cell surface Fc and C3 receptors, and might, therefore, be unresponsive to extracellular serum factors such as complement which stimulate intracellular killing. No population of neutrophils lacking CR3 complement receptors was detected. Furthermore, the expression of these receptors was enhanced on abscess neutrophils compared with peripheral blood neutrophils. FcγRII receptor expression was equivalent in the two populations. Extracellular serum was necessary for any abscess neutrophil bactericidal activity. To determine whether the bacteria within abscess neutrophils were a subpopulation selected for resistance to intracellular killing, we lysed the neutrophils and added the released bacteria to bactericidal assays with peritoneal exudate neutrophils. The bacteria were killed readily, but not as efficiently as log-phase organisms grown in vitro. As these studies have shown that neutrophils from abscesses are functional, with enhanced CR3 expression, and that bacteria persisting within abscesses are not resistant to intracellular killing, it is possible that the poor killing in vitro of bacteria engulfed in vivo by abscess neutrophils reflects depleted intracellulr killing mechanisms in those particular neutrophils, or results from the phagocytosis of organisms under conditions prevailing within abscesses, or is due to as yet undefined bacterial virulence factors.