Albumin has been used for volume resuscitation and supplementation in critically ill patients for over 50 years. While regarded as a "gold standard" colloid solution, albumin is associated with substantial cost, and questions have been raised about its safety and efficacy. A large-scale randomised controlled trial (the Saline vs. Albumin Fluid Evaluation [SAFE] study) demonstrated that albumin and saline were clinically equivalent treatments for intravascular volume resuscitation in a heterogenous population of critically ill patients. However, in patients with traumatic brain injury, albumin was associated with a significantly higher mortality and cannot be recommended for acute resuscitation of such patients. A potential beneficial role of albumin in patients with severe sepsis, particularly malaria, requires further study. Extrapolation of the results of the SAFE study to other, synthetic, colloid solutions requires caution, and a randomised controlled trial comparing albumin, starch and crystalloids in patients with severe sepsis is warranted. The safety of synthetic colloids in patients with traumatic brain injury should not be assumed. Although hypoalbuminaemia is associated with increased mortality, use of albumin for volume resuscitation of critically ill patients with a serum albumin concentration < or =25 g/L is not associated with reductions in mortality, duration of ICU stay or mechanical ventilation, or in use of renal replacement therapy. Similarly, there is no substantive evidence to justify the use of hyperoncotic albumin solutions for resuscitation or supplementation in critically ill patients. Albumin is a safe and effective resuscitation solution in critically ill patients without traumatic brain injury. However, the acquisition costs of albumin and synthetic colloids are more than those of crystalloids, and, as yet, colloids have not been proven to confer substantive benefits over crystalloids such as saline.