Rapid crack propagation can occur along pressurised pipelines of polyethylene, if a well-defined (but as yet unpredictable) critical pressure is exceeded. The standard for type-testing of pipe to meet exacting service conditions is currently set by full-scale tests. Results from the British Gas full-scale test method are presented and the effects of material, pipe diameter and wall thickness, extrusion process, pressurising fluid and temperature on the critical pressure are reviewed in the light of the Irwin-Corten "strain energy" analysis. These observations have guided the development of a new, complementary small-scale method. It is demonstrated that a small-scale technique designed to promote a steady state can, like the full-scale test, locate a precise transition in crack behaviour from prompt arrest to steady-state propagation, but that the effect of reducing the test size is to reduce the critical pressure at which this transition occurs. A correlation between the techniques is therefore necessary.