High throughput technologies like transcriptomics using DNA arrays or metabolomics employing a combination of gas chromatography with mass spectrometry provide valuable information about cellular processes. However, the measurements are often highly corrupted with noise of the experimental data which makes it sometimes difficult to draw reliable conclusions. Therefore, suitable statistical methods are needed for the evaluation of the experimental data to distinguish changes caused by biological phenomena from random variations due to noise. This paper introduces a likelihood ratio test to multiple metabolome measurements. The method was tested to differentiate differential metabolite compositions obtained from the pathogenic bacterium Pseudomonas aeruginosa grown under various environmental conditions.