The primary aim of the present study was to examine longitudinal models to determine the function that best describes developmental change in processing speed during childhood and adolescence. In one sample, children and adolescents (N= 503) were tested twice over an average interval of 2 years on two psychometric measures of processing speed: Visual Matching and Cross Out. In another sample, children and adolescents (N= 277) were tested four times, every 6 months, on Cross Out. Age-related changes in performance on both tasks were examined using six longitudinal models representing different hypotheses of growth. Linear, hyperbolic, inverse regression, and transition models yielded relatively poor fit to the data; the fit of the exponential and quadratic models was substantially better. The heuristic value of these latter models is discussed.