Left ventricular (LV) midwall geometry has been described conventionally as the sum of the chamber radius and half of the wall thickness; this convention is based on the assumption of uniform transmural thickening during systole. However, theoretical considerations and experimental data indicate that the inner half (inner shell) of the LV wall thickens more than the outer half (outer shell). Thus, an end-diastolic circumferential midwall fiber exhibits a relative migration toward the epicardium during systole. As a result, the conventional method provides an overestimate of the extent of the midwall fiber shortening. We developed an ellipsoidal model with a concentric two-shell geometry (nonuniform thickening) to assess midwall fiber length transients throughout the cardiac cycle. This modified midwall method was used in the analysis of LV cineangiograms from 15 patients with systemic arterial hypertension and 14 normal subjects. Study groups were classified according to LV mass index (LVMI): 14 normal subjects (group I), eight hypertensive patients with a normal LVMI (group II), and seven hypertensive patients with an increased LVMI (group III). There were no significant differences in LV end-diastolic pressure or volume among the three groups; the ejection fraction was slightly greater in group II (70 +/- 5%) than in groups I (65 +/- 8%) and III (66 +/- 4%), but this trend did not achieve statistical significance. Values for endocardial and conventional midwall fractional shortening (FS) were also similar in the three groups. By contrast, FS by the concentric two-shell geometry (modified midwall method) in group III (16 +/- 2%) was significantly less than that seen in groups I and II (21 +/- 4% and 21 +/- 5%, respectively; both p less than 0.05). This difference achieves greater importance when it is recognized that mean systolic circumferential stress was lower in group III (151 +/- 22 g/cm2) than in groups I and II (244 +/- 37 g/cm2 and 213 +/- 38 g/cm2, respectively; both p less than 0.01). The midwall stress-shortening coordinates in six of the seven group III patients were outside the 95% confidence limits for the normal (group I) subjects. Thus, despite a normal ejection fraction, systolic function is subnormal in hypertensive patients with LV hypertrophy. Chamber dynamics provide an overestimate of myocardial function, especially when LV wall thickness is increased. This is due to a relatively greater contribution of inner shell thickening in pressure-overload hypertrophy.