After two patients received bacterially contaminated platelet transfusions, a prospective surveillance program was instituted to perform Gram staining and microbiologic culturing of platelets at the time of transfusion. In 12 months, 3141 random-donor platelet pools (prepared from 14,481 units) and 2476 single-donor apheresis units were cultured. All single-donor apheresis units were sterile, but 6 (0.19%) of the random-donor pools were found to be bacterially contaminated, with 1 unit of 5 in the pool being the source in each case. Contaminants were Staphylococcus epidermidis (4 cases), Bacillus cereus (1), and Staphylococcus aureus (1) at counts of 0.5 x 10(2) to 10(11) colony-forming units per mL in platelet pools and 10(3) to 10(13) colony-forming units per mL in source units. The contamination rate for units transfused at < or = 4 days (1.8/10,000) was significantly lower than that at 5 days (11.9/10,000; p < 0.05), as was the magnitude of contamination (p < 0.05). Use of the pretransfusion Gram stain on 4- and 5-day-old platelet pools was 100 percent sensitive (4/4 true positives) and 99.93 percent specific (1 false positive) in detecting contaminated pools. These data define the extent and magnitude of platelet bacterial contamination and demonstrate the efficacy of the pretransfusion Gram stain on platelet units stored for 4 and 5 days in preventing the transfusion of heavily contaminated units. It is concluded that the risk of platelet contamination is related to the duration of component storage.(ABSTRACT TRUNCATED AT 250 WORDS)