Highly purified insulins offer the possibility of reducing insulin antibody levels and insulin requirement. Those likely to respond cannot be predicted on clinical grounds and a simple laboratory test is recommended for this purpose. This is based on insulin binding capacity (IBC) of plasma and has been used to follow a group of 47 patients over six months. 47 patients previously treated with British soluble and isophane insulins were changed to highly purified Leo Neutral and Leo Retard. 36 showed a reduction of insulin requirement and two groups could be identified. An IBC of greater than 40 muU/ml was associated in 94% with a reduction of insulin dose; if the IBC was less the response was unpredictable. The initial IBC was related to the initial insulin requirement and to the eventual percentage reduction. Serial measurements in patients with high initial IBC showed a steady fall. Measurement of the IBC is a simple investigation; if the level is greater than 40 muU/ml changing to a highly purified insulin is likely to be associated with reduction of insulin requirement.