Senior management and the union executive committee of a forest products company set an organizational goal to reduce theft from approximately a million dollars a year to zero. Salaried and hourly employees, selected at random, were interviewed regarding their outcome expectancies for honest and dishonest behavior. The responses were categorized within a 2 × 2 empathy box (honest/dishonest behavior vs. positive/negative outcome expectancies) to allow the organization's leadership to understand from the employee's perspective why there was so much theft. This information was subsequently used to alter employee outcome expectancies which, in turn, changed behavior. Theft dropped to near zero.