Systemic emboli released during total knee replacement have been implicated as a cause of peri-operative morbidity and neurological dysfunction. We undertook a prospective, double-blind, randomised study to compare the cardiac embolic load sustained during computer-assisted and conventional, intramedullary-aligned, total knee replacement, as measured by transoesophageal echocardiography. There were 26 consecutive procedures performed by a single surgeon at a single hospital. The embolic load was scored using the modified Mayo grading system for echogenic emboli. Fourteen patients undergoing computer-assisted total knee replacement had a mean embolic score of 4.89 (3 to 7) and 12 undergoing conventional total knee replacement had a mean embolic score of 6.15 (4 to 8) on release of the tourniquet. Comparison of the groups using a two-tailed t-test confirmed a highly significant difference (p = 0.004). This study demonstrates that computer-assisted knee replacement results in the release of significantly fewer systemic emboli than the conventional procedure using intramedullary alignment.