Shoplifting is one of the most common and costly crimes, yet little data exist to determine reliably characteristics of the typical shoplifter or the modus operandi of the crime. It is a crime that has most often been studied using official, secondary data provided by either retail security personnel or law enforcement officers. Reliability issues plague these official data. Continuing the “dark figure of crime” tradition, this study examines shoplifting by covert observation with a camera system installed in a typical suburban retail drug store. A standardized data template was used to record the demographic and behavioral characteristics of shoppers. Significant numbers of shoppers (8.5%) were observed shoplifting. Logistic regression analysis reveals that, while members of some demographic groups shoplifted more often than others, behavioral indicators carried far more predictive power. The methodology and findings are considered within the larger context of the law enforcement and “profiling” literatures.