A comparative study of lag phases and growth rates of healthy, stressed, and sublethally injured Escherichia coli O157 cells in 10 enrichment broths was performed. The evaluation of enrichment protocols was validated by different end point detection methods (two PCR and two combined capture-plate methods). Tryptic soy broth b [TSB (b)] provided the fastest growth (max = 1.00 1 0.06 h- ) but failed to recover oxidative-stressed E. coli O157. TSB (a), TSB-yeast extract medium, TSB supplemented with 8 mg/liter novobiocin plus 16 mg/liter vancomycin (TSB+), buffered peptone water (BPW), and BPW supplemented with 8 mg/liter vancomycin (BPW+V) enabled resuscitation of E. coli O157 cells independent from precultural conditions. Modified TSB plus 10 mg/liter novobiocin (mTSB+N), EC medium, EC reduced bile salts medium (ECred), TSB (b), and TSB supplemented with 8 mg/liter novobiocin plus 16 mg/liter vancomycin plus 2 mg/liter rifampin plus 1 mg/liter K-Telluriet plus 1.5 g/liter bile salts no. 3 (TSB++) all failed to recover E. coli O157 cells for at least one type of stress. The use of TSB (a), TSB+, BPW, and BPW+V was compared with that of mTSB+N (International Organization for Standardization reference broth) for reliable detection of low numbers of healthy, stressed, and sublethally injured E. coli O157 (approximately 10 CFU/10 g) from foods (sprouted seeds, fermented sausage, raw milk, and raw ground beef). When low numbers of healthy cells were inoculated, BPW, BPW+V, TSB, TSB+, and mTSB+N enabled growth until detectable numbers within 6 h of enrichment at 41.5 degrees C. Results showed that mTSB+N failed to recover to detectable numbers E. coli O157 cells sublethally injured by freeze and food stresses, in contrast to what was obtained with BPW and BPW+V. This study highlights that using mTSB+N for recovery of E. coli O157 from foods may yield false-negative results.