The processing-speed theory of cognitive aging proposed by Salthouse supposed that processing speed is the single mechanism which mediates cognitive aging. Measurements of processing speed involve a spectrum of speed measures from very simple perceptual-motor response tasks to high level executive functions. We supposed multiple processing speed training derived from modified speed measurements could broaden the training effects into a wider range of cognitive functions rather than restricted in very-near and near tasks. In study 1, 15 traditionally used processing speed tasks were analyzed and categorized into two categories according to results of factor analysis (perceptual-motor processing speed and central processing speed). In study 2, 41 elderly over 60 years old participated in 8 sessions of multiple processing speed training. Each participants joined either perceptual-motor processing speed training (PMPS) or central processing speed training (CPS). Processing speed performance of the elderly could be improved to the average level of the young adults' first performance after training, and PMPS abilities are easier to be improved compared to CPS abilities. Multiple processing speed training couldn't improve all the performance of processing speed, especially high level cognitive function.