We are developing a vendor-independent archive and on top of that a data warehouse for mass spectrometry metabolomics data. The archive schema resembles the communitydeveloped object model, the Java implementation of the model classes, and an editor (for both mzData XML files and the database) have been generated using the Eclipse Modeling Framework. Persistence is handled by the JDO2 -compliant framework JPOX. The main content of the Data Warehouse are the results of the signal processing and peak-picking tasks, carried out using the XCMS package from Bioconductor, putative identification and mass decomposition are added to the warehouse afterwards. We present the system architecture, current content, performance observations and describe the analysis tools on top of the warehouse. Availability: http://msbi.ipb-halle.de/