Informed recovery of the European eel (Anguilla anguilla) requires reliable information on stock dynamics and effects of management. No global European eel stock model is presently available, so demographic or extrapolation models are used to estimate eel production at local sub‐stock levels. Model inputs are often difficult to estimate and highly uncertain, which can affect accuracy of predictions. A sensitivity analysis was thus applied to the German Eel Model for the River Weser to evaluate how uncertainties in ten key input variables affected predicted eel escapement biomass. Uncertainty in the proportion of female eels and their age‐specific natural mortality, hydropower‐related mortality, and restocking had the largest influence on overall uncertainty. Conversely, uncertainty in eel predation by cormorants and age composition of immigrating male eels had small effects on outcome uncertainty. Based on expert judgment about the difficulty of parameter estimation, number of restocked eels, the proportion of females among outmigrants, and losses due to hydropower are the most feasible variables to address through monitoring. Better estimation of input variables and their uncertainties is crucial to improve eel stock assessment and management.