OBJECTIVE--To detect and characterise rapid temporal changes in the left ventricular response to exercise in patients with ischaemic heart disease and to relate these changes to the functional severity of coronary artery disease. BACKGROUND--The gamma camera does not allow the detection of rapid changes in cardiac function during exercise radionuclide ventriculography, the monitoring of which may improve the assessment of patients with ischaemic heart disease. METHODS--A miniature nuclear probe (Cardioscint) was used to monitor continuously left ventricular function during exercise in 31 patients who had coronary angiography for suspected coronary artery disease. A coronary angiographic jeopardy score was calculated for each patient. RESULTS--The coronary jeopardy score ranged from 0 to 12 (median 4). Ejection fraction fell significantly during exercise from 46% to 34%. Patients were divided into two groups based on the response of their ejection fraction to exercise. In 14 patients (group I), the peak change in ejection fraction coincided with the end of exercise, whereas in the other 17 patients (group II) the peak change in ejection fraction occurred before the end of exercise, resulting in a brief plateau. The peak change in ejection fraction and the time to its occurrence were independent predictors of coronary jeopardy (r = -0.59, p < 0.001 for peak change and r = -0.69, p < 0.001 for time to that change). The rate of change in ejection fraction was the strongest predictor of coronary jeopardy (r = -0.81, p < 0.001). In group I the peak change in ejection fraction was a poor predictor severity of coronary disease (r = -0.28, NS), whereas the time to peak and the rate of change in ejection fraction were good predictors (r = -0.65 and r = -0.73, p < 0.01). In group II the peak, the time to the peak, and the rate of change in ejection fraction were good predictors of coronary jeopardy (r = -0.75, r = -0.61, and r = -0.83, p < 0.01). CONCLUSION--The rate of change of ejection fraction during exercise can be assessed by continuous monitoring of left ventricular function with the nuclear probe, and is the best predictor of functionally significant coronary artery disease.