The purpose of the present experiment was to examine the principle of encoding specificity, manipulating recognition tasks. Two variables were operated. The first variable was types of frequency associates, high frequency associates and low frequency associates; respectively. The second variable was three types of recognition tasks. The subjects in Condition I were given free association tasks, having 30 extralist cue words. In Condition II, the subjects were given free associates prepared by the experimenter. In Condition III, free associates generated by the subjects in Condition I were presented to the subjects. The results consonant with the principle of encoding specificity revealed only in Condition I. Other two conditions showed recognition superiority over recall. These results were interpreted as inter-task interference.