Phrenic nerve injury and diaphragmatic dysfunction can be induced by cardiac operation. The clinical consequences are not well-established. We evaluated 13 consecutive patients over a 2-year period with unexplained and prolonged difficulties in weaning from mechanical ventilation. The mean time of measurement from the operation day was 31 +/- 19 days (range 8 to 78). With the same technique we also evaluated 12 control patients: four patients at day 1 after cardiac operation while they were still intubated; four normally convalescing patients at day 7 or 8 after cardiac operation; and four patients who required prolonged mechanical ventilation because of another identified cause after cardiac operation. Diaphragmatic function was evaluated at the bedside from esophageal and gastric pressure measurements. A low or negative ratio of gastric pressure swing to transdiaphragmatic pressure swing, indicative of diaphragm dysfunction, was found in all 13 patients (mean -0.39 +/- 0.64). The difference between the 13 patients and all control groups was found to be highly significant. Transdiaphragmatic pressure measured during a maximal voluntary inspiratory effort and transdiaphragmatic pressure measured during a short, sharp sniff were markedly diminished (28 +/- 18 cm H2O and 13 +/- 15 cm H2O, respectively) in the 13 patients, significantly different from values in the four control patients studied at day 7 or 8. Transdiaphragmatic pressure measured after magnetic stimulation in four patients was also markedly reduced (7 +/- 5 cm H2O) as compared with normal theoretic values. Aminophylline infusion had no effect on any of these parameters. In one of two patients evaluated a second time, about 5 weeks later, a marked improvement was observed. Estimating the prevalence of clinically relevant diaphragmatic dysfunction, we found it to be 0.5% when no topical cooling was used and 2.1% when iced slush with no insulation pad was added for myocardial protection (p < 0.005). The most striking finding was that the clinical course of the 13 patients was marked by severe intercurrent events, including cardiorespiratory arrest after early tracheal extubation in 5 patients, nosocomial pneumonia in 11, prolonged mechanical ventilation in all (58 +/- 41 days), and a fatal outcome in 3. We conclude that prolonged postoperative diaphragmatic dysfunction may cause severe life-threatening complications after cardiac operation and can be limited to some extent by avoiding the use of iced slush topical cooling of the heart.