Echocardiographic data from 92 younger normal subjects (1 month to 23 years of age) and 136 older normal subjects (20-97 years of age) were pooled and analyzed to obtain prediction equations for normal echocardiographic values. Using a bivariate regression model with the assumption that variability is constant as a percentage of the expected value, we developed regression equation and graphs that allow calculation of a 95% prediction interval for several echocardiographic measurements as a function of the subject's age and either body weight or body surface area. Body weight could be substituted for body surface area with no loss of precision. Further, examination of residuals showed that the linear prediction model fit well for all ages and all echocardiographic measurements studied. The measurements were obtained using the recently published standards recommended by the American Society of Echocardiography.