The aim was to assess if the reduction in polyethylene wear with highly crosslinked polyethylene suggested by studies with knee simulators is confirmed in patients with a knee arthroplasty. The use of a conventional or a highly crosslinked polyethylene was randomly assigned intraoperatively. Twelve months after surgery a knee arthrocentesis was performed and the synovial fluid of 17 patients in each group was studied analysing the number, size and shape of the polyethylene particles by scanning electron microscope. We found no significant differences in the concentration, size or morphology of polyethylene particles between groups. The great variability in the number of particles between individuals suggests that in vivo polyethylene wear depends on many factors and probably the type of polyethylene is not the most significant.