Age-specific distribution of clinical measurements in cross-sectional study is described in this paper. Since the distribution of measurements usually varies with age, a model with an agedependent structure is needed. We propose here a statistical method for describing the age-specific distribution using an extension of the power-normal-model. The age-dependent parameters are to be estimated through a nonparametric smoothing technique based on the local likelihood method. As a consequence, we can compute a smoothed percentile curve of measurements with reference to age. Several kinds of clinical measurements are analyzed to determine the proposed method.