The application of chemometric tools to analyze multi-way data has become popular, especially for excitation–emission matrix (EEM) fluorescence spectroscopy where PARAFAC often is employed to resolve the pure component profiles of mixtures of fluorophores. There are, however, some features in a typical EEM that do not follow the required low-rank tri-linear structure required for PARAFAC to work optimally. The most significant of these features is the light scatter effects that form diagonal lines in the landscapes. These cannot be modeled by one (or a few) factor(s) in the decomposition step, and has so far been removed by subtracting a standard, inserting missing values, or weighting these areas down. This paper suggests a novel method, which models the first order Rayleigh scatter as a separate set of factor(s) in the decomposition step by shifting the Rayleigh scatter into becoming low-rank bi-linear. This method is easier to implement than current methods and provides good results; both requirements for increasing the use of fluorescence spectroscopy by the non-expert user.