All available controlled studies of warm versus cold and antegrade versus retrograde delivery of cardioplegia were reviewed to assess the incidence of perioperative stroke and adverse neuropsychological outcomes. Nine randomized trials and substudies and two studies with immediate historical consecutive controls reported neurological outcomes and were described as warm versus cold. Pooled event rates for perioperative stroke were 1.5% for warm antegrade, 3.14% for warm retrograde, 1.7% for cold antegrade, and 0% to 1.2% for cold retrograde. Examining within trial differences, only one study showed a significant disadvantage to warm 4.5% versus cold 1.4% on incidence of perioperative stroke, but the design does not permit determination of whether the difference is due to systemic temperature, retrograde coronary perfusion, or other factors. Furthermore, if only warm (> 33 degrees C) versus cold (< 30 degrees C) systemic perfusion is examined in all studies for the incidence of stroke irrespective of cardioplegia temperature or antegrade versus retrograde coronary perfusion (warm 2.1%; cold 1.6%), the above study remains a significant outlier. This suggests that the differences found are unlikely to be due to temperature but may be related to antegrade versus retrograde coronary perfusion. Review of randomized trials evaluating neuropsychological function post-cardiopulmonary bypass (post-CPB) also failed to reveal any advantage related to temperature of systemic perfusion. Since manipulations that are most likely to give rise to cerebral embolization are uniformly carried out at normothermia at the beginning and end of the operation, it is not entirely unexpected that the incidence of neurological events was found to be independent of the temperature of CPB.(ABSTRACT TRUNCATED AT 250 WORDS)