Human intoxication or infection due to bacterial food contamination constitutes an economic challenge and a public health problem. Information on the in situ distribution and expression of pathogens responsible for this risk is to date lacking, largely because of technical bottlenecks in detecting signals from minority bacterial populations within a complex microbial and physicochemical ecosystem. We simulated the contamination of a real high-risk cheese with a natural food isolate of Staphylococcus aureus, an enterotoxin-producing pathogen responsible for food poisoning. To overcome the problem of a detection limit in a solid matrix, we chose to work with a fluorescent reporter (superfolder green fluorescent protein) that would allow spatiotemporal monitoring of S. aureus populations and targeted gene expression. The combination of complementary techniques revealed that S. aureus localizes preferentially on the cheese surface during ripening. Immunochemistry and confocal laser scanning microscopy enabled us to visualize, in a single image, dairy bacteria and pathogen populations, virulence gene expression, and the toxin produced. This procedure is readily applicable to other genes of interest, other bacteria, and different types of food matrices.