The marked decrease in neuropsychiatric morbidity (NPM) following coronary artery bypass (CAB) over a period of four years led us to carry out a retrospective study in order to identify the cause. Two hundred fifty-eight consecutive CAB procedures were performed between 1983 and 1986. For 133 patients (group A), the CAB procedure was performed with left ventricular vent (LVV) and for 125 patients (group B) without LVV. In group A, 93 patients (group A1) had cardiopulmonary bypass (CPB) with a bubble oxygenator and 40 patients (group A2) had CPB with a membrane oxygenator. In group B, 30 patients (group B1) had CPB with a membrane oxygenator without filter in the arterial line and 95 patients (group B2) had CPB with a membrane oxygenator and filter in the arterial line. The reduction in NPM arose from the removal of the LVV, as the incidence of NMP was in group A 24.6% versus 12.3% in group B (p < 0.05). The substitution of the bubble oxygenator by a membrane oxygenator does not alter the incidence of NPM: 23.3% in group A1 versus 27.5% in group A2 (p > 0.05) nor did incorporation of a filter in the arterial line: 10% in group B1 versus 13.1% in group B2 (p > 0.05). Introduction of air in the left ventricle via an LVV catheter opening and coronary arteriotomy therefore causes gazeous microembolic events responsible for NPM following CAB.