Studies were conducted to determine the relative effectiveness of whole soak [current Bacteriological Analytical Manual-(BAM) Salmonella method], quarter, stomach, and blend methods for the recovery of Salmonella organisms from internally and externally contaminated tomatoes. Tomatoes were subjected to three inoculation methods: surface inoculation, internal inoculation by injection, and immersion with single Salmonella serovars. The inoculation levels ranged from 1 to 100 CFU/tomato for surface and injection inoculation or 1 to 100 CFU/mL for immersion inoculation. Tomatoes were held for 3 days after inoculation at 2-6 degrees C prior to initiation of analysis. Contaminated tomatoes were soaked, quartered, stomached, and blended in appropriate portions of Universal Pre-enrichment broth, and incubated for 24 h at 35 +/- 2 degrees C. The BAM Salmonella culture method was followed thereafter, and tomatoes were treated as a low-microbial-load food. The stomaching procedure was significantly (P < 0.05) more effective than the whole soak procedure for recovery of internalized Salmonella from tomatoes (by injection). The blending procedure was arithmetically superior to the stomaching procedure for detection of internalized Salmonella from tomatoes (by immersion). The blending procedure showed the same effectiveness as the whole soak procedure for the detection of Salmonella on tomato surfaces. Comparisons between test portion-to-broth ratios (weight to volume) showed that a 1:3 test portion-to-broth ratio had a better buffering capacity for blended tomatoes than a 1:1 test portion-to-broth ratio. It is recommended that the current whole soak BAM tomato sample preparation procedure be replaced with a blending procedure and a 1:3 test portion-to-broth ratio.