The accuracy of digital subtraction angiography (DSA) for determination of left ventricular (LV) systolic wall thickness and muscle mass was evaluated in 20 patients (mean age 50 +/- 11 years). Conventional LV angiograms were digitized and subtracted using a combined subtraction mode ('mask mode' and 'time interval difference' subtraction). Wall thickness and muscle mass were determined at end-diastole, after the first- and second-third of systole and at end-systole. M-mode echocardiography (Echo), which was obtained from beam selection of the two-dimensional echocardiogram and conventional angiography (LVA), served as reference techniques. Angiographic LV wall thickness and muscle mass were determined according to the technique of Rackley in both, right (RAO) and left (LAO) anterior oblique projections, whereas echocardiographic wall thickness was measured just below the mitral valve orthogonal to the posterior wall (= LAO equivalent). Percent wall thickening was calculated in all patients. LV end-diastolic wall thickness and muscle mass correlated well between DSA and LVA (LV end-diastolic wall thickness in LAO projection r = 0.72, biplane LV end-diastolic muscle mass r = 0.83), LV end-systolic wall thickness (1.44 vs 1.33 cm, P less than 0.05) and percent wall thickening (52 vs 42%, P less than 0.05) compared favourably between echocardiography and DSA but was significantly larger when echocardiographically measured than with DSA (LAO projection). DSA and echocardiography showed a good correlation in regard to LV end-diastolic and end-systolic wall thickness (correlation coefficient r = 0.89, standard error of estimate SEE = 0.15 cm or 13% of the mean value).(ABSTRACT TRUNCATED AT 250 WORDS)