Abstract : We introduce in this paper LocFaults, a new flow-driven and constraint-based approach for error localization. The input is a faulty program for which a counter-example and a postcondition are provided. To identify helpful informa-tion for error location, we generate a constraint system for the paths of the control flow graph for which at most k con-ditional statements may be erroneous. Then, we calculate Minimal Correction Sets (MCS) of bounded size for each of these paths. The removal of one of these sets of constraints yields a maximal satisfiable subset, in other words, a max-imal subset of constraints satisfying the post condition. To compute the MCS, we extend the algorithm proposed by Liffiton and Sakallah  in order to handle programs with numerical statements more efficiently. The main advantage of this flow-driven approach is that the computed sets of suspicious instructions are small, each of them being asso-ciated with an identified path. Moreover, the constraint-programming based framework of LocFaults allows mix-ing Boolean and numerical constraints in an efficient and straightforward way. Preliminary experiments are quite en-couraging. * This work was partially supported by ANR VAC-SIM (ANR-11-INSE-0004), ANR AEOLUS (ANR-10-SEGI-0013), and OSEO ISI PAJERO projects.