Similarity search has received great attention on modern database applications involving complex objects, since the queries over such objects are seldom based on exact matches, but rather on some notion of similarity, specific to each domain. However, the SQL query language does not provide support for similarity queries. This paper proposes to include similarity queries to SQL, adding a powerful set of similarity operators. Two kinds of complex objects are considered: those monolithically stored as Binary Large Objects, and those stored as sets of attributes in a relation. Concerning monolithic objects, we regard specificallysimilarity search in image datasets. The paper also proposes ways to efficiently store and retrieve complex objects, showing a prototype developed to validate the concepts and the syntax presented.