A review of the studies using 50% human serum as a diluent for the serum bactericidal test has shown correlations with patient outcome. Human serum used as diluent of the patient's serum appears to be essential because of high protein binding of some antibiotics. An inoculum of 10(5)-10(6) bacteria/ml and a bactericidal criteria of 99.9% killing are technical aspects that have gained popularity. Careful timing of serum collection for the assay is important. Neither the macrotube nor microtiter techniques are entirely satisfactory. The latter method, however, has the advantage of being more reproducible than the macrotube method, less cumbersome and requiring less serum. Preliminary guidelines for performing and interpreting the test are provided. Future research should be directed toward making the microtiter technique more sensitive for identifying antibiotic tolerance, developing effective methods to eliminate the need for human serum as a diluent and obtaining more clinical correlations.