The Beddoes/Leishman dynamic-stall model has become one of the most popular for the provision of unsteady aerofoil data embedded in much larger codes. The underlying modeling philosophy was that it should be based on the best understanding, or description, of the associated physical phenomena. Even though the model was guided by the flow physics, it requires significant empirical inputs in the form of measured coefficients and constants. Beddoes provided these for a Mach number range of 0.3-0.8. This paper considers one such input for a Mach number of 0.12, where, from the Glasgow data, it is shown that the current stall-onset criterion, and subsequent adjustments, yield problematic results. A new stall criterion is proposed and developed in the best traditions of the model. It is shown to be very capable of reconstructing the Glasgow's data for stall onset both the ramp-up and oscillatory tests.