In this work, we used barley leaves suffering from a stress, for measurements of chlorophyll a fluorescence with an imaging fluorometer. We compared selected fluorescence parameters (FP) determined from the measurements of control (no stress) and afterwards stressed sample by classical statistical comparison (Mann-Whitney test) and by statistical comparison of shapes of distributions of the FPs (two-sample Smirnov test). We have found that there exist examples where statistically significant difference is not revealed using the classical statistical comparison (for given critical level), but statistically significant difference is revealed using comparisons of distributions (for the same critical level). It implies that the shape of statistical distribution of a FP is more sensitive to a stress of a sample than median of the FP. Further, the comparison of changes in shapes of statistical distributions of FPs is therefore more suitable for early detection of plant stress than a classical statistical comparison. The observed changes in the distributions of FPs are discussed.