The MB/BacT ALERT 3D System (MB/BacT) (Organon Teknika, Boxtel, The Netherlands) is a fully automated, nonradiometric system with a revised antibiotic supplement kit designed for the recovery of mycobacteria from clinical specimens. In a multicenter study, the recovery rate of acid-fast bacilli (AFB) and the mean time to their detection from clinical specimens was determined by using the MB/BacT system. Data were compared to those assessed by the radiometric BACTEC 460 system (B460) and by culture on Löwenstein-Jensen (L-J) solid medium. A total of 2,859 respiratory and extrapulmonary specimens were processed by the N-acetyl-L-cysteine (NALC)-NaOH method using two different concentrations of sodium hydroxide; 1.5% was adopted in study design A (1,766 specimens), and 1.0% was used in study design B (1,093 specimens). The contamination rates for MB/BacT were 4.6% (study design A) and 7.1% (study design B). One hundred seventy-nine mycobacterial isolates were detected by study design A, with 148 Mycobacterium tuberculosis complex (MTB) isolates and 31 nontuberculous mycobacteria (NTM) isolates. Overall recovery rates were 78.8% for MB/BacT (P = 0.0049), 64.2% for L-J (P < 0.0001), and 87.1% for B460, whereas they were 84.5, 70.9, and 91.2%, respectively, for MTB alone. A total of 125 mycobacteria were detected by study design B, with 46 MTB and 79 NTM. Overall recovery rates by the individual systems were 57.6% (P = 0.0002), 56.8% (P = 0.0001), and 80% for MB/BacT, L-J, and B460, respectively, whereas the rates were 91.3, 78.3, and 97.8% for MTB alone. By study design A, the mean times to detection of smear-positive MTB, smear-negative MTB, and NTM were 11.5, 19.9, and 19.6 days, respectively, with the MB/BacT; 8.3, 16.8, and 16.6 days, respectively, with the B460; and 20.6, 32.1, and 27.8 days, respectively, with L-J medium. By study design B, the mean times were 15.1, 26.7, and 26 days with the MB/BacT; 11.7, 21.3, and 24.8 days with the B460; and 20.4, 28.7, and 28.4 days with L-J medium. Identification was attempted by probing (Accuprobe) MB/BacT-positive bottles within the first working day following instrument positive flag. Results were compared to those obtained in the B460 positive vials by the p-nitro--acetylamino--hydroxypropiophenone (NAP) test (study design A) or by the Accuprobe assay (study design B). About 90% of MTB and 100% of NTM could be identified, showing turnaround times closely related to those obtained by combining B460 and the NAP test or the Accuprobe assay. In conclusion, even though recovery rates were shown to be lower than B460, especially for NTM, and contaminants were somewhat higher, MB/BacT represents a valuable alternative to the radiometric system, especially in those laboratories where disposal of radioactive waste is restricted. Finally, when AFB are cultured in nonradiometric liquid media, our data (detection times and bacterial overgrowth rates) suggest that decontamination with 1.5% NaOH may be more suitable than the standard NALC-NaOH.