This paper reports experimental data obtained on an instrumented screw extruder. The material used was a starch powder that was independently characterised in a uniaxial compaction experiment. The data have been carefully analysed at some length and are compared with a simple model based on an enhancement of the well-known model by Darnell and Mol (SPE J. 12 (1956) 20). The Darnell and Mol model is modified in order to extend its predictive capacity to that required by experiments which have been conducted with an industrial-sized powder extruder. Not only are the predictions shown to be in good accord with the experiments, but the model also contributes to the understanding of the extrusion process by identifying the variables and parameters to which the system is most sensitive, and thereby becomes a potentially powerful design tool. This insight is gained at the expense of the additional detail which might have been obtained with traditional finite element methods but which would have fallen beyond the experimental resolution of the system. Indeed, the experimental data currently available do not allow discrimination between the low-order model reported here and a model of much greater sophistication. In addition, and more importantly perhaps, the paper demonstrates that even in the apparently diametrically opposed cases of, on the one hand, traction-arrested flow (as exemplified by uniaxial compaction) and, on the other, traction-induced flow (in a single-screw extruder), it is possible to abstract from the simpler system the principles, methods and data that can be applied to the more complex geometry.