We studied the relationship between victimization and fear of crime longitudinally, analyzing data from the Observatory of the North-West (Italian national sample, N51,701, two waves: January 2006 and January 2007). We modeled fear of crime at T2 using as independent variables: (a) the main sociodemographic variables and fear of crime, as assessed at T1; (b) direct victimization; and (c) indirect victimization. Recent direct victimization was the most effective victimization predictor of both concrete and abstract fears, followed by multiple or repeat direct victimization. On the other hand, direct victimization occurring in the 12 months before the first wave did not influence fear. Recent indirect victimization and, above all, multiple or repeat indirect victimization influenced concrete, but not abstract, fear of crime, while direct victimization occurring in the 12 months before the first wave did not foster fear. We discuss the limits of this work and possible further directions.