This study compared the effects of self-evaluation and adult-dispensed tokens on room-cleaning behavior of six predelinquent youths in a residential token-economy setting. The self-evaluation token system proved to be as effective as the external adult-administered system in increasing room-cleaning behavior and was more effective in maintaining performance after contingencies were withdrawn. The self-evaluation token system increased resistance to extinction, compared to the external token system, and appeared to be a useful component for a traditional token-economy system.