We investigated an outbreak of 396 Salmonella enterica serotype I 4,5,12:i:- infections to determine the source. After 7 weeks of extensive hypothesis-generation interviews, no refined hypothesis was formed. Nevertheless, a case-control study was initiated. Subsequently, an iterative hypothesis-generation approach used by a single interviewing team identified brand A not-ready-to-eat frozen pot pies as a likely vehicle. The case-control study, modified to assess this new hypothesis, along with product testing indicated that the turkey variety of pot pies was responsible. Review of product labels identified inconsistent language regarding preparation, and the cooking instructions included undefined microwave wattage categories. Surveys found that most patients did not follow the product's cooking instructions and did not know their oven's wattage. The manufacturer voluntarily recalled pot pies and improved the product's cooking instructions. This investigation highlights the value of careful hypothesis-generation and the risks posed by frozen not-ready-to-eat microwavable foods.