The 2017 Grenfell Tower fire spread rapidly around the combustible facade system on the outside of the building, killing 72 people. We used a range of micro- and bench-scale methods to understand the fire behaviour of different types of facade product, including those used on the Tower, in order to explain the speed, ferocity and lethality of the fire. Compared to the least flammable panels, polyethylene-aluminium composites showed 55x greater peak heat release rates (pHRR) and 70x greater total heat release (THR), while widely-used high-pressure laminate panels showed 25x greater pHRR and 115x greater THR. Compared to the least combustible insulation products, polyisocyanurate foam showed 16x greater pHRR and 35x greater THR, while phenolic foam showed 9x greater pHRR and 48x greater THR. A few burning drips of polyethylene from the panelling are enough to ignite the foam insulation, providing a novel explanation for rapid flame-spread within the facade. Smoke from polyisocyanurates was 15x, and phenolics 5x more toxic than from mineral wool insulation. 1 kg of burning polyisocyanurate insulation is sufficient to fill a 50m3 room with an incapacitating and ultimately lethal effluent. Simple, additive models are proposed, which provide the same rank order as BS8414 large-scale regulatory tests.