Although caffeine attenuates the vasodilatation produced by dipyridamole and adenosine, and is therefore contraindicated when these agents are used for myocardial perfusion scintigraphy, caffeine levels in clinical patients undergoing standard imaging protocols have not been studied. Eight-six patients undergoing clinically indicated intravenous dipyridamole (n = 75) or adenosine (n = 11) thallium-201 myocardial perfusion scintigraphy, all of whom reported abstention from products containing caffeine for 24 h, were studied prospectively. Blood samples were drawn prior to initiation of the pharmacologic infusion, and serum caffeine levels were determined using an enzyme immunoassay technique. Results of these determinations were correlated with maximum pulse and blood pressure changes measured during and immediately after the stressor infusion, and thallium imaging findings. Detectable caffeine levels were found in 34 patients (40%), ranging from 0.1 to 5.0 mg/l. There was no significant difference in mean systolic blood pressure decrease or mean pulse increase between patients with caffeine levels > 1.0 mg/l (20.4 +/- 18.2 mmHg, 11.0 +/- 8.9 BPM; n = 5) and those with lower (0.1 to 0.9 mg/l) (15.4 +/- 9.5 mmHg, 14.4 +/- 8.2 BPM; n = 29) or no detectable caffeine levels (18.0 +/- 11.5 mmHg, 16.6 +/- 10.1 BPM; n = 52). Redistribution on thallium imaging was also identified with a similar frequency in these three groups (2/5, 40%; 8/29, 28%; 22/52, 42% respectively).