Background Colloids are widely used in the replacement of fluid volume. However doubts remain as to which colloid is best. Different colloids vary in their molecular weight and therefore in the length of time they remain in the circulatory system. Because of this and their other characteristics, they may differ in their safety and efficacy. Objectives To compare the effects of different colloid solutions in patients thought to need volume replacement. Search strategy We searched the Cochrane Injuries Group specialised register, the Cochrane Controlled Trials Register (2002 Issue 3), MEDLINE (1994-2002/07), EMBASE (1974-2002 August week 1), and the National Research Register (2002 issue 3). Bibliographies of trials retrieved were searched, and drug companies manufacturing colloids were contacted for information. The search was last updated in September 2002. Selection criteria Randomised and quasi-randomised trials comparing colloid solutions in critically ill and surgical patients thought to need volume replacement. The main outcomes measured were death, amount of whole blood transfused, and incidence of adverse reactions. Data collection and analysis Two authors independently extracted the data and assessed the quality of the trials. Main results Fifty-seven trials met the inclusion criteria, with a total of 3659 participants. Quality of allocation concealment was judged to be adequate in 20 trials and poor or uncertain in 37. Deaths were obtained from 36 trials. For albumin or PPF versus hydroxyethyl starch (HES) 20 trials (n=1029) reported mortality. The pooled relative risk (RR) was 1.17 (95% CI 0.91, 1.50). For albumin or PPF versus gelatin four trials (n=542) reported mortality. The RR was 0.99 (0.69, 1.42). For gelatin vs HES 11 trials (n=945) reported mortality, RR was 1.00 (0.78,1.28). RR was not estimable in the albumin vs dextran, gelatin vs dextran, and HES vs dextran groups. Thirty-six trials recorded the amount of blood transfused, however quantitative analysis was not possible due to skewness and variable reporting. Fifteen trials recorded adverse reactions, but none occurred. Authors' conclusions From this review, there is no evidence that one colloid solution is more effective or safe than any other, although the confidence intervals are wide and do not exclude clinically significant differences between colloids. Larger trials of fluid therapy are needed if clinically significant differences in mortality are to be detected or excluded.