Less is generally known about the ideas, events and personalities which drove developments permitting the evolution of haemodialysis as a clinically useful form of palliation and treatment, than its subsequent success and failures. This pre-history of haemodialysis is summarized here. One must remember that with hindsight we can now discern connections between ideas and developments which were not perceptible in their time, and that progress towards any new idea, material or piece of hardware was usually random and undirected, and outcomes uncertain. We must also remember the many blind alleys we can now safely ignore, to give a spurious continuity to the development of ideas. The prehistory of dialysis begins with study of the diffusion of solute and solvent in osmosis in living systems and experimental settings, and the retention of potentially toxic substances in kidney failure, during the 18th and early 19th centuries. These two areas came together in work in the mid-19th century on diffusion of gases and liquids, and showed that natural and synthetic membranes could selectively hinder different solutes. This explained osmosis and allowed semi-permeable membranes to be used and designed. These ideas underpinned the subsequent history of both dialysis using body cavities such as the peritoneum (not discussed here) and ex vivo dialysis of blood. To perform this, new membranes and anticoagulants were needed. These led to the first attempts in animals in 1912-3, and human patients in 1924-8, but only the purification and synthesis of newer materials such as cellulose and heparin allowed practical and successful haemodialysis to evolve in the 1940s.