This article presents an integrated and systematic investigation into the interaction and effects of the abrasive flow machining factors, through its three major process variables of the media velocity, temperature and quantity, which play an essential role in interfacing the ‘machine’ with ‘workpiece’ and ‘media’ corners of the abrasive flow machining triangle. The article also presents predictive models, main effects and industrially useful rule-of-thumb tools. Collectively, these variables offer machine operators the ability to manipulate the process behaviour when the opportunity to modify geometry and media levels is unavailable. In the media and geometry corners of the triangle, capital expenditure is required to adjust levels, making them economically and temporally limiting. The machine adjustments are physically limited by the hardware in use; however, this research finds that the range of response magnitude can vary significantly among the three process outcomes studied, that is, surface roughness, material removal and peak height reduction. Using a standard media and testpiece set-up, data are collected using a 3³ full factorial experiment design and translated into a response surface design (Box–Behnken) for predictive model development. Application to oil and gas industry parts is shown whereby data are utilised to aid in the abrasive flow machining of production parts.