The purpose of this work was (i) to investigate the feasibility of a previously reported upstream processing method for PCR template preparation to facilitate the detection of Escherichia coli O157:H7 from ground beef and (ii) to assess the impact of cell growth (no growth in the matrix versus growth in the matrix) on molecular detection limits. Two food matrices (autoclaved and raw ground beef) were evaluated in all studies. For no-growth experiments, 10-g meat samples were inoculated with 10(2) to 10(7) CFU/g E. coli O157:H7 and then homogenized. The homogenates were processed to remove large particulates and inhibitors using a two-phase upstream processing method consisting of two sequential centrifugation steps, the second of which used titanous hydroxide to facilitate bacterial immobilization. After upstream processing, sample concentrates were extracted for DNA isolation and amplified by PCR. For growth experiments, 10-g meat samples were inoculated at 1 CFU of E. coli O157:H7 per gram, allowed to grow to 10(2) to 10(7) CFU/g, and then processed for PCR assay. Cell recoveries after upstream processing ranged from 15.9 to 77.6% and were not facilitated by the use of titanous hydroxide, as compared with a saline control (P > 0.05). Bacterial cell recovery and PCR detection limits were similar when comparing autoclaved ground beef and raw ground beef, but cell recoveries were highly variable for raw ground beef samples in which E. coli O157:H7 cells were allowed to grow before processing for detection. Overall, PCR detection limits approximated 10(3) CFU/g of ground beef for all treatments. These results indicate that use of model food systems may not always provide an accurate replication of real-world conditions when evaluating PCR detection limits.