During operations for total joint replacement done in operating rooms with conventional ventilation the mean air contamination varied considerably among the 15 hospitals studied. The range was from 51 to as many as 539 bacteria-carrying particles per cubic metre. When the data from all the hospitals were grouped according to the mean level of bacterial airborne contamination, including operations done in control and in ultraclean air, there was a good correlation between the air contamination and the joint sepsis rate. There was also a correlation between the mean values of air contamination and the numbers of bacteria isolated from wound wash-out samples; but the apparent efficiency of the sampling method varied a great deal among the hospitals carrying out this procedure. From this data it would seem that by far the largest proportion of bacteria found in the wound after the prosthesis had been inserted reached it by the airborne route. With the mean air contamination found in the control series, 164 bacteria-carrying particles per cubic metre, this proportion was as much as 95 per cent. The risk of joint sepsis varied widely among the 19 hospitals. The differences between the highest and lowest being probably as much as 20-fold. However, the effect of an ultraclean air environment was asimilar at all hospitals.