During May and June of 2005, 26 persons in several states were infected by a single strain (isolates indistinguishable by pulsed-field gel electrophoresis) of Salmonella enterica serotype Typhimurium after eating cake batter ice cream. The cake mix used to prepare the cake batter in the ice cream was implicated by epidemiologic investigation as the source of Salmonella contamination. Initial tests did not detect Salmonella in cake mix collected during the outbreak investigation. The objective of this study was to evaluate different procedures to isolate Salmonella from the implicated cake mix, cake, and ice cream. All outbreak-associated food samples (14 samples) were collected during the outbreak investigation by health departments of several of the states involved. Different combinations of Salmonella isolation procedures, including sample size, preenrichment broth, enrichment broth, enrichment temperature, and isolation medium, were used. Salmonella Typhimurium was isolated from two cake mix samples; the food isolates were indistinguishable from the outbreak pattern by pulsed-field gel electrophoresis subtyping. Universal preenrichment broth was substantially better than was lactose broth for preenrichment, and tetrathionate broth was better than was Rappaport-Vassiliadis broth for isolating Salmonella from the two positive cake mix samples. Although more typical Salmonella colonies were observed on plates from enrichment cultures grown at 35°C, more confirmed Salmonella isolates were obtained from plates of enrichment cultures grown at 42°C. Brilliant green agar, xylose lysine tergitol 4 agar, xylose lysine desoxycholate agar, Hektoen enteric agar, and bismuth sulfite agar plates were equally effective in isolating Salmonella from cake mix. The best combination of preenrichment-enrichment conditions for isolating the outbreak strain of Salmonella was preenrichment of cake mix samples in universal preenrichment broth at 35°C for 24 h, followed by enrichment in tetrathionate broth at 42°C for 24 h.