Abstract Objectives. We compared three techniques to predict functional recovery after revascularization. Background. Recently, fluorine-18 (F-18) fluorodeoxyglucose in combination with single-photon emission computed tomography (SPECT) has been proposed to identify viable myocardium. Thallium-201 reinjection and low dose dobutamine echocardiography are used routinely for this purpose. Methods. Seventeen patients (mean [±SD] left ventricular ejection fraction 36 ± 11%) were studied. Regional and global ventricular function were evaluated before and 3 months after revascularization by echocardiography and radionuclide ventriculography, respectively. Myocardial F-18 fluorodeoxyglucose uptake (during hyperinsulinemic glucose clamping) was compared with rest perfusion assessed with early thallium-201 SPECT. On a separate day, low dose dobutamine echocardiography and post-stress thallium-201 reinjection SPECT were simultaneously performed. Results. The sensitivities for F-18 fluorodeoxyglucose/thallium-201, thallium-201 reinjection and low dose dobutamine echocardiography to assess recovery were 89%, 93% and 85%, respectively; specificities were 77%, 43% and 63%, respectively. Stepwise logistic regression indicated that F-18 fluorodeoxyglucose/thallium-201 was the best predictor. In hypokinetic segments, the combination of F-18 fluorodeoxyglucose/thallium-201 and low dose dobutamine echocardiography was the best predictor. Global function improved (left ventricular ejection fraction increased > 5%) in 6 patients and remained unchanged in 11. All three techniques correctly identified five of six patients with improvement. Fluorine-18 fluorodeoxyglucose/thallium-201 identified all patients without improvement; low dose dobutamine echocardiography identified 9 of 11 without improvement; and thallium-201 reinjection identified 6 of 11 patients without improvement. Conclusions. Fluorine-18 fluorodeoxyglucose/thallium-201 SPECT was superior to the other techniques in assessing functional recovery. Integration of metabolic and functional data is necessary, particularly in hypokinesia, for optimal prediction of improvement of regional function.