Conventional observation of a fermentation process by using NAD(P)H‐dependent fluorescence provides a mean for cell monitoring, but its practical application has been very limited due to many factors that affect culture fluorescence. In this work we studied correlation of scanning fluorometry signals with various process variables and with cell metabolic states and showed that scanning fluorometry is more informative than conventional NAD(P)H‐dependent fluorescence. A stepwise multiple‐regression procedure has been used to select meaningful signals from the whole spectrum, and we found that with three or four components of the excitation−emission plane the process variables can be portrayed with a rather good accuracy. The approach is demonstrated on the examples of wild‐type or recombinant Escherichia coli and Saccharomyces cerevisiae fermentations.