A probabilistic approach can help to predict the buckling loads of cylindrical shells, where spatial varying imperfections are often modeled as random fields. The shape of the fields depends strongly on the autocorrelation structure. Underlying uncertainties like a small sample size or imprecise measurements make it practically impossible to define a crisp correlation function. In this paper, the classical probabilistic approach is therefore extended to a fuzzy stochastic approach in context of design of cylindrical shells. More exactly, the polymorphic uncertainty model (fp-r) in Graf et al. (2015) is used to take into account natural variability and incompleteness. The applicability of these uncertainty model is demonstrated with real measured geometric imperfections from Arbocz and Abramovich (1979). Due to the small sample size of tested shells in the data bank, the correlation parameters are defined as fuzzy input variables. Consequently, the results are stability loads presented as fuzzy output variables aiming to consider aleatory and epistemic uncertainties in a decision making process.