We randomly allocated 50 total knee replacements to scrub teams wearing body-exhaust suits (BES) or Rotecno occlusive clothing. The effectiveness of the clothing was assessed using air andwound bacterial counts. Bacteria were recovered from 62% of wounds (64% BES, 60% Rotecno). The mean air count was 0.5 CFU/ m3 with BES and 1.0 CFU/m3 with Rotecno (p = 0.014). The mean wound counts were 14 bacteria/wound with BES and eight bacteria/wound with Rotecno (p = 0.171). There was no correlation between the air and wound counts (r = -0.011, Spearman's). The higher air counts suggest that Rotecno occlusive clothing is less effective than BES, but wounds were equally contaminated with both types of clothing suggesting that at very low levels of air contamination the contribution of bacteria to the wound from the air is irrelevant. Even doubling the air counts from 0.5 to 1.0 CFU/m3 had no detectable effect on the wound. This allows a reassessment to be made of other sources of contamination the effect of which would previously have been overwhelmed by contamination from air.