The growth of two strains of Serratia marcescens in blood components was tested in this study. One of the strains had been implicated in the epidemic of transfusion-associated sepsis experienced in Denmark and Sweden in 1991. In whole blood with a final concentration of 100 colony- forming units per mL of S. marcescens, there was an immediate reduction of more than 95 percent of colony-forming units, but no reduction of the bacterial concentration if the blood had been white cell-reduced before inoculation. This is interpreted as an effect due to phagocytosis by white cells and as a lack of bactericidal effect of the plasma. A reduction to 10 percent of the original concentration, observed if the blood had a nominal content of white cells, was most likely due to phagocytosis. White cell reduction by filtration after inoculation further reduced the bacterial concentration of one of the strains tested, but, after a 1-week lag phase, growth accelerated to high concentrations by 6 weeks. In platelet-rich plasma prepared from S. marcescens-inoculated units, abundant growth was found after 24 hours, increasing to very high concentrations (10(12) colony-forming units/mL) during 10-day storage at 22 +/− 2 degrees C. Keeping the whole blood at ambient temperature for 20 hours before preparation of platelet-rich plasma caused only temporary reduction of bacterial concentration in the S. marcescens experiments, but resulted in a complete absence of bacteria in the platelet-rich plasma for 10 days in control experiments performed with Staphylococcus epidermidis. In another control experiment using Staphylococcus aureus, the initial growth pattern was similar to that of S. marcescens, but the final bacterial concentrations were not as high. Thus, the protection against bacterial growth was insufficient with respect to S. marcescens as well as to S. aureus. Growth of one of two strains of S. marcescens was inhibited in whole blood. This growth inhibition did not occur in white cell-reduced units. The resistance of S. marcescens to antibacterial mechanisms in the blood shows that contamination, even with low numbers of these bacteria, may be dangerous.