We assessed the reliability and validity of a formula, based on echocardiographically derived parameters, to calculate left ventricular mass index (LVMI) in a group of 15 chronic hemodialysis patients. All patients had M-mode echocardiography before and after a hemodialysis session. Echocardiograms were interpreted by 2 observers blind to each other's measurements. Interobserver reliability for LVMI was high (r = 0.94, p < 0.0007). LVMI decreased in 11 of 15 patients during dialysis and increased in 4. The mean difference in LVMI between pre- and posthemodialysis was 26.2 +/- 15 g/m2 (p < 0.0001). End-diastolic diameter decreased from 53.5 +/- 5.9 to 49.5 +/- 7.5 mm (p = 0.0016). These data indicate that measurement of LVMI is highly reproducible in hemodialysis patients but that it changes significantly over the course of a hemodialysis session. Its use as an outcome measure in clinical trials in hemodialysis patients should be interpreted with caution.