Nationally representative health examination surveys have collected anthropometric measurements on children and adolescents in the U.S. These data may serve as very crude indicators of total fat or fat-free body mass, but do not yield information on body composition per se. Attempts to accurately describe trends in obesity for children and adolescents have been complicated by a number of limitations associated with body measurements taken over time in sequential surveys. Inferences from limited anthropometric measures should be made cautiously. Factors to consider include the selection of cutpoints for trend analysis, changes that may be within range of technical measurement error, and use of data at extreme percentiles. Ostensible trends may be genuine, but they may also be influenced by methodological differences across surveys. Comparative data are presented for selected body measures from 1963 to 1980 for U.S. children and adolescents sampled in national health examination surveys, and apparent differences are discussed.