Two experiments were carried out to investigate the validity of hypotheses regarding the encoding difficulty effects on a free recall. Subjects were required to judge which of two targets held a stronger associative relationship with a paired presented word followed by recognition and free recall tests. Two types of paired targets were used in Experiment 1: AA paired targets (associate and associate) and AN paired targets (associate and non-associate). Positive correlation was observed between a free recall and the mean clustering size. However, neither positive nor negative correlation was observed between a free recall and a false recognition as an index of elaboration. Three types of paired targets were used in Experiment 2: AA, AN and NN paired targets (non-associate and non-associate). AA paired targets were more recalled than AN and NN ones. Mean clustering size was highest in AA paired targets followed by AN and NN ones. These results were interpreted as showing that the organization hypothesis was adequate in explaining the encoding difficulty effects on an incidental free recall.