The Biosoil project carried under the scope of the Forest Focus EC regulation aims to constitute a European-wide database in order to monitor forest soil condition, and feed the Soil Profile Analytical Database of Europe (SPADE-M) with measured soil profiles. The final database will gather information on approximately 5000 plots dealing with soil classification, profile description and soil sample analysis over 2 survey rounds. This huge amount of data of different nature and of different origin has to be stored in a single database in order to be manipulated, retrieved for analysis and shared. Besides, soil related databases have to face two critical issues: the need to store properly not only the data but also the method used to obtain it; and the ability to check the data for errors, as many inputs are entered manually. With these considerations in mind, we have designed a generic software architecture that comprises: (1) the central relational database that favours generic tables because they are more suitable for analytical menu changes; (2) a metadata base that describes data formats, data units and codes and analytical methods; (3) a validation database that stores checking SQL queries that generate warnings when data seems suspicious, and also a record of such data and the follow up of its validation so as to assure a quality assurance/quality control (QA/QC) procedure. Furthermore this allows testing the validation relevance. (4) Data input is made via a web application that betters the interaction between data input and database managers. The system described here manages the data workflow from its initial spreadsheet format to data checked for errors and arranged in a self-descriptive database that will be stored by the European Commission Joint Research Centre. Concepts developed may be generalised to any soil database to better data quality management and assessment.