Thirty-four hip and knee total arthroplasty failures due to infection, loosening, intractable pain and dislocation were examined for synovial and capsular tissue debris. Simiquantitative assessments were made for metal polyethylene, acrylic and cellular debris and inflammation. The most important finding was that in the amounts normally shed into joints, debris particles do not appear to stimulate a polymorphonuclear response. The magnitude of this response correlates best with clinical and/or bacteriological evidence of infection. The quantity of acrylic particles could be correlated with clinical evidence of loosening. Since the correlation between infection and polymorphonuclear leukocytes, debris particles, and the reasons for failure is high, frozen section may aid the surgeon in determining the methods and timing of revision operations.