Flexible querying of information systems allows expressing complex preferences in user queries. Such preferences can be modeled by fuzzy bipolar conditions which are made of constraints c and wishes w and interpreted as "to satisfy c and if possible to satisfy w". We define in this article the main elements of the Bipolar SQLf language, which is an SQL-like querying language based on a bipolar relational algebra [11,3]. This language is an extension of the SQLf language [2,1]. Basic statements (projection, selection, etc.) are firstly defined in terms of syntax, evaluation and calibration. Then, complex statements, such as bipolar queries based on nesting operators are studied in terms of expression, evaluation, query equivalence and backward compatibility with the SQLf language.