This paper describes a potentially serious problem regarding medical force feedback micromanipulator. Medical micromanipulators are now actively applied to stomach cancer tests and in fields such as brain surgery operations, etc. However, this application has a very difficult challenge pertaining to the stability of the man—machine system itself. Basically, need to be extremely safe, because that is one of a very serious nature in our healthcare environment. In particular, an abnormal self-excited oscillation of medical force feed-back micromanipulator is considered to be a very serious problem. In this study, an abnormal self-excited oscillation means that the slave gripper oscillates without any relation to the movement of master gripper. Furthermore, this is a concern that generally occurs due to interaction between the characteristics of the finger nerve-musculoskeletal system of the operator and the bilateral servo system. For instance, “dead-time” of the finger nerve-musculoskeletal system is one of the concerned characteristics. In this paper, firstly we propose a new model of the nerve-musculoskeletal system of an operator based on a physiological study for keeping repeatability, because characteristics of the nerve-musculoskeletal system of an operator are not repeatable phenomenon in man—machine system. That is why the paper focuses on a new model of the human's finger used in the design of the micromanipulator. Afterwards the cause of the self-excited oscillation of the bilateral servo system was investigated in connection with dead-time of the finger nerve-musculoskeletal system of the operator. As the result, the self-excited oscillation observed by experiments was obtained through the simulations by using a totally new overall model. Finally, it has been confirmed that the characteristics of the finger nerve-musculoskeletal system of operator affects the stability of the overall systems at least. This new insight had been expected intuitively in previous studies and this study newly showed the objective model-based evidence for the insight.