An alternative to blood transfusion, based on oxygen-carrying solutions, has been sought for over a century. The present 'first-generation' haemoglobin-products were based on observations that crosslinking with, for example, glutaraldehyde, overcame subunit dissociation and renal toxicity. Experience with these solutions has shown that they can be vasoactive, sometimes increasing blood pressure, sometimes decreasing tissue perfusion and sometimes both. Clinical trials have been disappointing because of unexpected toxicity. The 'second-generation' products are based on a better understanding of the mechanisms of this vasoconstriction. Such products may seem counterintuitive by traditional standards, but it is hoped that they will be less toxic, more beneficial to patients, and more economical to produce.