There continue to be many arguments about the psychological mechanism of item method directed forgetting. The traditional data analysis methods used in previous studies were also difficult to quantify and estimate the storage and retrieval process behind the directed forgetting effect. Besides, the test in a series of memory processing models was limited to recall test and could not fully reflect all aspects of forgetting and memory retrieval. Therefore, based on the previous studies, the present study first constructed a new multinomial processing tree model of memory process used single word as learning material and comprehensively investigated the memory retrieval processed in the test stage phase. After the model was validated, this study manipulated the relevant variables of the theories to explore the internal psychological mechanism of item-method directed forgetting. The storage-retrieval model was used as a template to construct a multinomial processing tree model based on the free-recall-then-recognition paradigm, and the model was continuously improved. The confirmatory experiment conducted a 2 (coding depth: 1000 ms vs. 5000 ms) × 2 (learning-test interval: 0 min vs. 6 min) mixed design, and 32 college students participated in this experiment. The results showed that the model could well reflect the changes in the storage and retrieval of information during the memory process. Moreover, model assumptions were also supported, and the model could be applied to subsequent studies. To investigate the internal psychological mechanisms of item-method directed forgetting, the experiment conducted a 2 (group: forgetting group vs. remember group) × 2 (item type: TBF/R vs. TBR/R) × 2 (SOA: 1600 ms vs. 2600 ms) mixed design to examine the internal psychological mechanism of directed forgetting under item-method paradigm and 38 college students participated in this experiment. The results of the experiment showed that for the TBF/R items, the storage and retrieval capacity of the forgetting group were weaker than those of the memory group; for the TBR/R items, only the storage capacity of the forgetting group was stronger than that of the memory group. The participants were less able to retrieve TBF items than TBR items in the forgetting group. Moreover, for both short-term SOA and long-term SOA, the reaction time of TBF items that followed by a probe stimulus in the forgetting group was longer than that of the memory group. In conclusion, the multinomial processing tree model based on the free-recall-then-recognition paradigm can well reflect the changes of critical parameters by experimental manipulation. Therefore, it can be applied to the research of directed forgetting internal psychological mechanism. The internal psychological mechanism of the item-method directed forgetting is the result of the combined effect of selective rehearsal, attentional inhibition and retrieval inhibition. Furthermore, the cost of directed forgetting relied on the combined effect of attentional inhibition and retrieval inhibition, and the benefit depended on the combined effect of selective rehearsal and attention inhibition.