Abstract A pilot scale pelletizing system (capacity: 300 kg/h) with integrated die temperature control was built and evaluated using reed canary grass as a model raw material. A pelletizing series was performed using a full factorial design having independent factors such as die temperature (35–65 °C), raw material moisture content (12–15% w.b.), and raw material temperature/steam conditioning (0–2 kg/h), with continuous or discontinuous feed layer formation as responses. Die temperature was the most influential parameter on feed layer formation and was negatively correlated to continuous feed layer formation. An immediate response, changing discontinuous production patterns into continuous was experienced when die temperature was lowered by the temperature control system. Further, a partial least squares discriminant analysis model, using the experimentally varied parameters, predicted continuous or discontinuous feed layer formation with 100% accuracy. Thus, die temperature control can be considered as a fast and efficient method for overcoming feed layer losses in pelletizing.