For assessment of total knee arthroplasty outcome, various scoring systems have been introduced. The current study assessed the interobserver correlation of four commonly used total knee arthroplasty outcome scores. One hundred eighteen total knee arthroplasties were investigated by two independent observers, using the Hungerford score, the Hospital for Special Surgery score, the Knee Society score, and the Bristol score. Each score consisted of three subscores: pain, knee, and function. For the highest interobserver correlation was computed for the Bristol score (interobserver correlation coefficient, 0.88). For knee range of motion, flexion contracture, and extension lag there was high interobserver correlation (interobserver correlation coefficient > 0.8 each). For walking distance and walking aids, there also was a high interobserver correlation (interobserver correlation coefficient > 0.7 each). For clinical assessment of total knee arthroplasty, pain should be measured on a four-step system, the knee should be assessed by measurement of range of motion, extension lag, and flexion contracture, and function should be measured on a separate score assessing walking distance and walking aids.