The detection of Brucella bacteremia by subculture does not always correlate with a positive signal in the BACTEC NR730 nonradiometric system (Becton Dickinson Diagnostic Instrument Systems, Sparks, Md.). The effect of the inoculum size, pH, sodium polyanetholesulfonate, carbon sources (i-erythritol, sodium pyruvate, monosodium glutamate, D-glucose, and L-alanine), and urea in the release of CO2 was evaluated by using the reference strain Brucella melitensis 16M. In standard NR6 vials with or without blood, inocula 5 to 10 times larger (at least 265 CFU per vial) than those usually found in the blood of patients with brucellosis were necessary to produce a positive growth value (GV) in 4 days or less, and similar results were obtained with vials supplemented with the substrates listed above. GVs were consistently lower in vials with sodium polyanetholesulfonate than in vials without this agent. Vials with no blood inoculated with 265 CFU per vial showed turbidity 1 day before GVs became positive, proving that the major limiting detection factor was the low level of release of CO2 and not an inadequate growth medium. In NR6 vials buffered to pH 6.2, GVs became positive faster and were higher than those in standard vials. NR6 vials at pH 6.2 with 0.3% sodium pyruvate yielded a positive GV in the first day of bacterial turbidity.