The purpose of the present study was to compare conventional colony forming unit (CFU) enumeration with a bioluminescent measurement in order to quantitate living bacteria in BCG vaccine. Forty batches of BCG vaccine quantitated with respect to weight (10 each of concentration 120 mg/vial, 30 mg/vial, 1.5 mg/vial and 0.75 mg/vial, respectively) were tested. Adenosine triphosphate (ATP) was extracted using the boiling Tris-EDTA extraction method preceded by treatment with the ATP-hydrolysing enzyme apyrase, and the liberated ATP was assayed using a LKB 1251 luminometer. The sensitivity of the assay was 10(-11) M ATP which corresponded to 5 x 10(5) CFU/ml. A proportional relation between CFU and ATP content per vial was found up to 30 mg/vial, but for 120 mg/vial the CFU method seemed unreliable in contrast to the bioluminescence method. The measurements error for the logarithm (base 10) of ATP was 0.051 and this method was therefore more exact than the CFU method (0.097). Because the bioluminescence method is sensitive, reliable, time-saving and less expensive it seems preferable to CFU enumeration in the quality control of BCG vaccines.