The thermal fast pyrolysis of high density polyethylene (HDPE) has been carried out in a conical spouted bed reactor in the 450-715 degrees C range, and individual products have been monitored with the aim of obtaining kinetic data for the design and simulation of this process at large scale. Kinetic schemes have been proposed in order to explain both the results obtained in the laboratory plant and those obtained in the literature by other authors operating at laboratory and larger scale. Discrimination has been carried out based on the contribution of the variance of model parameters (stepwise regression) to the total variance explained by the model. The models based on that of Westerhout et al. [R.W.J. Westerhout, J. Waanders, W.P.M. Van Swaaij, Recycling of polyethene and polypropene in a novel bench-scale rotating cone reactor by high-temperature pyrolysis. Ind. Eng. Chem. Res. 37 (6) (1998) 2293-2300] do not adequately predict the experimental results, especially those corresponding to aromatics and char, which is probably due to the very short residence times attained in the conical spouted bed and, consequently, to the lower yields of aromatics and char. The model of best fit is the one where polyethylene degrades to give gas, liquid (oil) and wax fractions. Furthermore, the latter undergoes secondary reactions to give liquid and aromatics, which in turn produce more char.