We examined the changes in left ventricular (LV) end-diastolic volume (EDV) during symptom-limited upright exercise in 160 patients with coronary artery disease. Patients were divided into three groups based on the degree of exercise-induced LV dilation: group I (n = 66) had no increase in EDV (less than 10% change), group II (n = 38) had mild increase (10% to 24%), and group III (n = 56) had moderate or severe increase (greater than or equal to 25%). The patients with exercise-induced LV dilation had smaller resting EDV (p = 0.0001), higher resting LV ejection fraction (EF) (p = 0.008), and a greater decrease in EF with exercise (delta EF) (p = 0.007). A significant correlation was found between resting EDV and the degree of LV dilation with exercise (r = -0.45, p less than 0.001), and a weaker correlation between the degree of LV dilation and delta EF (r = -0.28, p less than 0.01). Multivariate discriminant analysis of important clinical, anatomic, and exercise descriptors identified three variables that best predicted the presence or absence of LV dilation during exercise: resting EDV (F = 38.2), resting EF (F = 24.2), and delta EF (F = 21.5). Thus, the degree of LV dilation during upright exercise is not related to age, gender, and extent of coronary artery disease, but is determined by the resting EDV, the resting EF, and the EF response to exercise.