Exercise thallium imaging is effective and provides information on exercise tolerance and symptoms, but is less sensitive if exercise is submaximal. When dipyridamole alone is used for such patients, the exercise parameters are lost and high background uptake results. This study examines dipyridamole combined with maximal exercise, to capitalize on the strengths of both techniques. Thallium tomography was performed in 122 patients with dipyridamole (group 1), exercise (group 2), or dipyridamole with maximum exercise (group 3). In patients with coronary artery disease, thallium tomography was abnormal in 92, 93 and 95% (P = ns), with reversible ischaemia in 82, 81 and 84% (P = ns) in the three groups, respectively. Among patients in groups 2 and 3 who exercised submaximally, detection of individual arterial stenoses was improved in group 3 (P < 0.05) and larger defects were found (2.4 versus 4.2 segments, P < 0.01). In group 3, no serious complications occurred and fewer vasodilatory side effects were seen than in group 1 (P < 0.01). The heart to splanchnic count ratio in groups 2 and 3 (1.61 versus 1.65, P = ns) was significantly greater than in group 1 (1.28, P < 0.0001). Compared with exercise alone, additional dipyridamole improved diagnostic accuracy in patients with limited exercise potential. Compared with dipyridamole alone, additional exercise provided useful exercise information, reduced vasodilatory side effects and improved scan quality. Combined dipyridamole and exercise stress is safe, and could be used routinely, but a larger study is required to establish if there is an overall diagnostic improvement.