Systemic approaches such as metabolomics are increasingly needed to improve the development of novel drugs. In this paper, we suggest a new strategy based on infrared spectroscopy which probes the global chemical composition of a sample. Seven cell lines from three tumour types were investigated and exposed to four classical anticancer drugs belonging to two classes characterized by a unique mechanism. First, each cell line was considered separately and a hierarchical clustering was built for the seven cell lines. Spectra clustered according to the drug mechanism of action for all the cell lines tested. Second, the similarities among drug mechanism spectral fingerprints were investigated for all the cell lines simultaneously. Difference spectra (the mean spectrum of the corresponding untreated cell line was subtracted) were computed so that the particular contribution of every cell line was eliminated and only the drug-induced differences could be compared. The hierarchical clustering shows a clear tendency to distinguish the two modes of action, revealing a very similar type of response to molecules with a similar mechanism. High throughput systems with 96-well plates are now available and a well established bioassay could be developed in order to provide an objective classifier for potential anticancer drugs.