Rhamnolipid is one of the most effective and commonly used biosurfactant with wide industrial applications. Systematic strategies were applied to improve rhamnolipid (RL) production with a newly isolated indigenous strain Pseudomonas aeruginosa EM I originating from an oil-contaminated site located in southern Taiwan. Seven carbon substrates and four nitrogen sources were examined for their effects on RL production. In addition, the effect of carbon to nitrogen (C/N) ratio on RL production was also studied. Single-factor experiments show that the most favorable carbon sources for RL production were glucose and glycerol (both at 40 g/L), giving a RL yield of 7.5 and 4.9 g/L, respectively. Meanwhile, sodium nitrate appeared to be the preferable nitrogen source, resulting in a RL production of 8.6 g/L. Using NaNO3 as the nitrogen source, an optimal C/N ratio of 26 and 52 was obtained for glucose- and glycerol-based culture, respectively. To further optimize the composition of fermentation medium, twenty experiments were designed by response surface methodology ((RSM) to explore the favorable concentration of three critical components in the medium (i.e., glucose, glycerol, and NaNO3). The RSM analysis gave an optimal concentration of 30.5, 18. 1, and 4.9 g/L for glucose, glycerol, and NaNO3, respectively, predicting a maximum RL yield of 12.6 g/L, which is 47% higher than the best yield (8.6 g/L) obtained from preliminary selection tests and single factor experiments (glucose and NaNO3 as the carbon and nitrogen source). The NMR and mass spectrometry analysis show that the purified RL product contained L-rhamnosyl-beta-hydroxydecanoyl-beta-hydroxydecanoate (RL1) and L-rhamnosy L-rhamnosyl-phydroxydecanoyl-beta-hydroxydecanoate (RL2). Meanwhile, HPLC analysis indicates that the molar ratio of RL1 and RL2 in the purified rhamnolipid product was ca. 1:1. (c) 2007 Elsevier Ltd. All rights reserved.