The in vitro effect that the presence of components of non-specific immunity (serum plus polymorphonuclear neutrophils) has on the bactericidal activity of co-amoxiclav was explored against Streptococcus pneumoniae strains exhibiting an amoxicillin MIC > or =4 mg/L. Eight penicillin-resistant clinical isolates non-susceptible to co-amoxiclav with MICs of 4 (two strains), 8 (four strains) and 16 mg/L (two strains) were used. Values of MBC were identical to MIC values in all cases. Time-kill curves were performed with co-amoxiclav concentrations achievable in serum after a single oral dose administration of the new 2000/125 mg sustained-release formulation. Results were expressed as percentage of reduction of initial inocula after 3 h incubation. Control curves showed growth with no reduction of initial inocula. Against strains with MIC of 4 and 8 mg/L, the results obtained with the antibiotic alone or with the presence of factors of non-specific immunity were similar, with a weak combined effect due to the intrinsic activity of co-amoxiclav (reductions of initial inocula ranging from 70 to 99.16%). Against strains with MIC of 16 mg/L, the addition of PMN in the presence of serum increased the reduction of bacterial load provided by the aminopenicillin, even at sub-inhibitory concentrations (25.8% versus 51.1% at 0.5 x MIC concentration--8/0.5 mg/L). This combined activity against strains with an amoxicillin MIC of 16 mg/L which decreased the bacterial load may be important in preventing bacterial proliferation within the host and the transmission of resistant clones to others.