Left ventricular (LV) wall thickness and muscle mass are important measures of LV hypertrophy. In 24 patients LV end-diastolic wall thickness and muscle mass were determined (two observers) by digital subtraction angiocardiography (DSA) and conventional LV angiocardiography (LVA). Wall thickness was determined over the anterolateral wall of the left ventricle according to the technique of Rackley (method 1) or by planimetry (method 2). Seventeen patients were studied at rest and seven during dynamic exercise. Wall thickness correlated well between LVA and DSA; the best correlations were obtained by a combined subtraction mode using either method 1 or 2 (method 1, r greater than or equal to 0.80; method 2, r greater than or equal to 0.75). The standard error of estimate of the mean (SEE) was slightly lower for method 2 (less than or equal to 10%) than for method 1 (less than or equal to 13%). DSA significantly overestimated wall thickness by 5-7% with method 1 and underestimated by 12-14% with method 2. Muscle mass correlated well between LVA and DSA; the SEE was less than or equal to 15% for method 1 and less than or equal to 12% for method 2. Overestimation of muscle mass by DSA was 7-11% with method 1 and underestimation was 13-15% with method 2. It is concluded that LV wall thickness can be determined accurately by DSA with an SEE ranging between 10 and 13%. Determination of LV muscle mass is slightly less accurate and the SEE is slightly larger ranging between 13 to 17%. With method 1, wall thickness and muscle mass were overestimated and with method 2 underestimated.