Changes in transfusion behaviour induce a widespread use of colloid plasma substitutes, the range of which has recently been enlarged by the marketing of starch derivatives. The product chosen depends, at least in part, on its adverse effects, anaphylactoid reactions being a part of these. This study aimed to discover the frequency and severity of these reactions according to the type of substitute available in France, to look for possible risk factors, and determine the mechanisms involved. A prospective inquiry was carried out in 49 public and private hospitals spread throughout France. It lasted for 15 months, between June 1991 and October 1992. A data sheet was filled in for each patient who was given a plasma substitute, whether or not there was an incident. When a reaction did occur, an assessment was carried out in two stages: straightaway, with the measurement of the concentrations of serum tryptase, antigelatin antibodies, urinary methylhistamine; and four to six weeks later, with skin tests. A series of 19.593 patients was thus collected: 48.1% were given gelatins, 26.7% starches, 15.7% albumin, and 9.5% dextrans. 43 anaphylactoid reactions were recorded, giving an overall frequency of 0.219%, or 1 reaction for 456 patients. The frequency differed according to the substitute considered: 0.345% for gelatins, 0.273% for dextrans, 0.099% for albumin, and 0.058% for starches. These reactions (grades III and IV) were serious in 20% of cases. Multivariate analysis revealed four independent risk factors: giving gelatins (odds ratio: 4.81), giving dextrans (odds ratio: 3.83), a history of drug allergy (odds ratio: 3.16), and being male (odds ratio: 1.98). The relative risks of anaphylactoid reactions due to one type of substitute with respect to another were estimated to be 6 times less for starches with respect to gelatins, and 4.7 times less than with dextrans. The relative risk of albumin is 3.4 times less than that of gelatins, and almost identical to that of the starches. An immuno-allergological assessment was only carried out in 15 patients who had been given a gelatin (Plasmion). IgE-dependent anaphylaxis was proven in 7 of these. To conclude, it was shown that gelatins and dextrans should be avoided in patients with a known history of drug allergy. When a reaction does occur, an allergological assessment must be carried out, as this may be due to specific antibodies. Should this prove to be the case, that particular substitute would be contraindicated for the rest of the patient's life.