We have compared our standard stress protocol (adenosine combined with exercise) with the new stress agent arbutamine, for thallium-201 myocardial perfusion imaging (MPI) in order to assess the comparative value of arbutamine. We studied 23 patients referred for MPI, and each patient had two studies (18 males, median age 66 years, five with previous myocardial infarction). Uptake scores were assigned to each of nine segments, and the extent and severity of defects were measured using a polar plot. Haemodynamic changes were greater with arbutamine (rate-pressure product increase 78% vs 51%, P = 0.003). Symptoms were experienced by 21 patients with arbutamine and 16 with adenosine (P = 0.07). Agreement between the techniques for classification of patients as normal or as having reversible, fixed or mixed defects was good (19 of 23 studies, 83%, kappa = 0.76). Agreement for similar classification of segments was also good (82%, kappa = 0.71). Segmental agreement for stress scores was good (86%, kappa = 0.77). However, mean size of stress defect was larger with adenosine (83+/-52 pixels vs 65+/-48 pixels, P<0.05), though severity and reversibility were similar (P = NS). We conclude that arbutamine provides comparable results to those obtained with adenosine and exercise and that the observed differences are not clinically significant.