This paper presents an impact evaluation of the French enterprise zone program which was initiated in 1997 to help unemployed workers find employment by granting a significant wage-tax exemption (about one third of total labor costs) to firms hiring at least 20% of their labor force locally. Drawing from a unique geo-referenced dataset of unemployment spells in the Paris region over an extensive period of time (1993-2003), we are able to measure the direct effect of the program on unemployment duration, distinguishing between short- and medium-term effects. This is done by implementing an original two-stage empirical strategy using individual data in the first stage and aggregate data and conditional linear matching techniques in the second stage. We show that although the enterprise zones program tended to "pick winners", it is likely to be cost-ineffective. It had a small but significant effect on the rate at which unemployed workers find a job (which is increased by a modest 3 percent). This effect is localized and significant only in the short run (i.e. at best during the 3 years that follow the start of the policy).