Boneless beef roasts (longissimus muscles) were vacuum-packaged and then the bags were injected with one of six gas mixtures: 100% O2, 20% CO2 +80% N2, 50% CO2 + 50% O2, 20% CO2 +80% O2, 25% CO2 + 25% O2 +50% N2 or 51% CO2 + 30% O2 + 18% N2 + 1% CO. One group of roasts, vacuum-packaged without added gas served as controls. Roasts were stored for 0–35 days at 1–3 C. At five weekly intervals, steaks were removed from roasts in each treatment and examined after storage for 5 days under retail display conditions. Psychrotrophic plate counts of roasts stored in modified gas atmospheres were usually higher than those stored in conventional vacuum packages. Differences in lactobacillus counts between roasts stored in modified gas atmospheres and those stored in vacuum packages rarely were statistically significant. Counts of retail steaks prepared from roasts stored in various gas atmospheres were usually slightly higher than those prepared from comparable vacuum packaged roasts. In most instances these differences were not statistically significant. Initially, the microbial flora of vacuum-packaged beef roasts consisted primarily of Moraxella-Acinetobacter and Pseudomonas spp. Lactobacilli predominated on roasts at later storage intervals even on roasts stored in atmospheres initially containing 100% O2 or 20% CO2 + 80% O2. Pseudomonas spp. remained a substantial part of the microflora of roasts stored in high O2 containing atmospheres.