A crossover study to compare the effects of seven different dialysers on intradialytic symptoms in 37 patients during dialysis with acetate-containing dialysate was performed at five centres in four countries. The same manufacturing lot of each dialyser and of blood line sets were used by all centres. The same clinical data (duration of dialysis, blood pressure, weights, temperature, drugs, symptoms, and treatments) and technical data (blood flow, dialyser clearance, and ultrafiltration rate) were collected. Kt/V for urea was used to determine dialysis prescribed. Intradialytic symptoms and signs were measured hourly or when observed by staff using the haemodialysis treatment form (see Introduction). After each week of treatment with a particular dialyser, patients completed a questionnaire relating to the presence and severity of symptoms. (Only presence or absence of symptoms are presented.) Wide differences in dialysis duration and blood flow between centres were noted. These may have contributed to the differences between centres in relationship to staff reported responses to different dialyser: Dialysers with the lowest incidence of both signs and symptoms and of chest pain, back pain, and itching (arbitrarily designated bioincompatibility symptoms) were the Duo-Flux and Filtral, with the G120 M, the CD 4000, and the T 150 having the highest incidence. By patient questionnaire the most biocompatible dialysers were the T 150, F 60, and the Filtral, with the most symptom producing being the G120 M and the G10-3N. Perceptions of symptoms between patients and staff differed substantially overall and between centres. Hypersensitivity reactions were noted in two patients, both occurring with cuprammonium cellulose hollow-fibre dialysis, despite adherence to manufacturers' instructions concerning saline priming and removal. Both patients showed antibody titres greater than 1:160 against ethylene oxide-HSA. Ethylene oxide was not detected (limit of detection 1 part per million) in dialysers, blood line sets, or fistula needles. The study suggests that dialysis symptom reporting is complicated by individual perceptions, staff reactions, and the efficiency of recording. In this study ethnic and cultural differences must be added to the haemodynamic differences and other prescription-related elements in influencing symptoms. Despite these problems a hierarchy of dialyser-related symptoms and signs could be discerned which largely paralleled laboratory findings of biocompatibility. Future comparative studies relating symptomatology to membrane and dialyser structure should consider the variables identified as influencing symptoms and their reporting.