The extraction of functional dependencies is a fundamental activity in the database design recovery process. Existing algorithms for this task are computationally expensive and appear to be impractical if applied to large legacy database instances, e.g., their performance deteriorates when number of attributes or/and instances is large. This paper presents strategies for parallelising the functional dependencies discovery process. We propose three parallel discovery models which are based on horizontal, vertical, and matrix database table slicing techniques. We exploit both program parallelism and data parallelism in our implementations. The results are discovery approaches that are more applicable to large real world databases.