A prospective randomised clinical study comparing the functional performance and biocompatibility of a new cellulose diacetate variant (Dicea) in which the degree of hydroxyl group substitution differs, with cellulose diacetate and low flux polysulfone incorporated into commercially produced hollow fiber hemodialysers with a surface area 1.5-1.6 m2 has been undertaken. All dialysers studied demonstrated clinically acceptable performance in terms of their small molecular removal characteristics, with minor statistical but not clinical differences. Use of both cellulose diacetate membranes but not low flux polysulfone resulted in a reduction in plasma beta(2) microglobulin levels. The membranes were impermeable to albumin, but showed some permeability to low molecular weight proteins. The average protein recovery from the dialysis fluid was 3105 mg for Dicea, 2913 mg for cellulose diacetate and 2842 mg for low flux polysulfone. For Dicea the white cell count by 15 minutes had declined to 68% of pre treatment value, compared with 59% and 86% for cellulose diacetate and low flux polysulfone. The differences between Dicea and cellulose diacetate were not significant, but both cellulose based membranes differed from low flux polysulfone (p = 0.0015). There was a strong evidence of differences between the membranes in respect of C5a and C5b-9 generation (p = 0.0001) but not for C3a (p = 0.16) furthermore the levels of C5b-9 generated during dialysis also showed a significant positive correlation compared to C5a for all membranes. (Pearson's correlation coefficient = 0.856, p = 0.0001). It is concluded that the two cellulose diacetate membranes are not identical, with the differences observed being a consequence of the degree of acetyl substitution, resulting in alteration of membrane structure and the method of sterilization. The clinical significance of these differences are difficult to characterize but the modification of the cellulose structure appears to be a promising method to improve the biocompatibility of cellulose membranes. The improved biocompatibility offered by this method still falls short of that achieved with low flux synthetic membranes such as Fresenius Polysulfone.