In this study, serial reproducibility was determined on standard M-mode (M-mode) and two dimensional guided M-mode (2-D M-mode) records for ten normal volunteers and 20 patients. Overall intraobserver variability in the normal group was 518 +/- 1.8 percent (M-mode) and 3.1 +/- 0.8 percent (2-D M-mode), and in the patient group 3.1 +/- 0.6 percent (M-mode) and 4.7 +/- 0.7 percent (2-D M-mode) (P = NS). Variability on serial examination, however, was twofold to threefold greater. In the normal group, reproducibility varied by 10.8 +/- 1.6 percent (M-mode) and 9.4 +/- 1.2 percent (2-D M-mode), while in the patient group, it was 8.7 +/- 0.9 percent (M-mode) and 9.4 +/- 0.7 percent (2-D M-mode). The lowest serial variability achieved was for the diastolic LV dimension on serial M-mode examination (4 percent); the largest variability pertained to measurement of the interventricular septum (14 percent). Other structures had intermediate variation in reproducibility. This study emphasizes that establishing performance variability for an echocardiographic laboratory is an important part of the interpretation of serial records.