Social networking communities have become an important communications platform, but the popularity of these communities has also made them targets for a new breed of social spammers. Unfortunately, little is known about these social spammers, their level of sophistication, or their strategies and tactics. Thus, in this paper, we provide the first characterization of social spammers and their behaviors. Concretely, we make two contributions: (1) we introduce social honeypots for tracking and monitoring social spam, and (2) we report the results of an analysis performed on spam data that was harvested by our social honeypots. Based on our analysis, we find that the behaviors of social spammers exhibit recognizable temporal and geographic patterns and that social spam content contains various distinguishing characteristics. These results are quite promising and suggest that our analysis techniques may be used to automatically identify social spam.