Maintaining mussels in good eating condition is critical to supplying live mussels through the distribution chain. This study assessed the impact of varying headspace gas composition on quality of modified atmosphere packaged (MAP) live mussels stored at 4˚C for 15 days. Three levels of MAP headspace oxygen gas percentage (MAP100, MAP80 and MAP60) were investigated in this study. Commercially packed (SB) samples were used as the control. Shelf-life was based on total viable count (TVC), odour and presence of spoilage bacteria, while meat condition, pH, total ammonia nitrogen (TAN in pouch water) and drip loss were used as quality indices. Mean TVC of pouch water and mussel meat in SB and experimental packs was just below microbial limit (6 log CFU mL-1) at day 0, but exceeded microbial limit by days 4 and 7 respectively. Mean headspace oxygen of MAP100 was 97% on day 0, but in SB it was 58%. By day 10, mean headspace oxygen in SB had decreased to 4.50 ± 2.66% whereas in MAP100 it was 83.80 ± 4.35%. MAP80 and MAP60 were intermediate at 70.77 ± 2.97% and 50.47 ± 2.20%. Hydrogen - sulphide producing bacteria were observed in all the samples from day 4, but were too numerous to count (TNTC) in the SB packs. Objectionable sulphide odour was noted in SB and MAP60 packs on day 7, but was not evident in MAP100 and MAP80 packs until day 10. Mean TAN exceeded the tolerance level (5 mg L-1) by day 4 in both MAP100 and SB packs. In contrast, mean TAN in MAP 80 and MAP 60 packs did not achieve this until day 10, which suggests lower metabolic rates in the intermediate MAP treatments. MAP80, which had the lowest mortality, is considered to be the most suitable oxygen composition to maintain shelf-life.