The American Society of Anesthesiologists Physical Status Classification System (ASA) ranks patients for risk of adverse events during a surgical procedure. The ASA classification is used as a surrogate for the patient's underlying severity of illness and has been recommended for use in Surgical Site Infection (SSI) and risk stratification. We assessed the predictive power of the ASA score for total knee replacement surgery infection, and compared it to a comorbidity score. All patients who had TKA (total knee arthroplasty) surgery performed during the period of 1993 to 1999 at one institution were identified. One hundred and thirteen infected cases were matched with 236 controls and nominal variables were statistically processed. A total co-morbidity score (TCOMORBID) was created to help the analysis. All possible predictors of infection were tested against infection in bivariate analysis. The association of the ASA score with infection was examined in detail. An ASA score beyond 2 showed an increased risk of infection. The average ASA score for the infected TKA group was 2.3 +/- 0.6, and the non-infected TKA average score was 2.6 +/- 0.7 (cohort effect). The relationship between the ASA score and TCOMORBID score was poor; Spearman rank correlation rho = 0.2, (p < 0.0001). In fact, the ASA score predicted only 6% of the occurrences of infection, but since it predicts 98% of the cases where there is no infection correctly, it is 70% accurate over all. Infection in TKA surgery was associated with an increased ASA score, but only when the high ASA score was due to a combination of specific co-morbidities. We propose that the ASA score should be cross-checked with the current co-morbidities, like rheumatoid arthritis or active infections in order to assess TKA infection risk.