Needle–tissue interactive force is vital for developing a virtual simulation and planning system (SPS) and optimizing needle control strategy, wherein the friction coefficient is an important parameter but difficult to pre-determine. The existing studies on friction coefficient estimation were lack of qualitative analysis and did not consider the effect of normal pressure and interactive velocity, which may lead to the inaccuracy of the friction calculation. In this paper, we proposed a novel semiempirical friction coefficient model based on the modified classical elastic friction theory that constructs the relationship between the friction coefficient and parameters such as the normal pressure and velocity. The proposed friction coefficient model is validated by using the computational inverse technique based on coupled finite element material point (CFEMP) contact algorithm. The results show that the friction coefficient between the needle and polyvinyl alcohol (PVA) tissue phantom varies from 0.091 to 0.242 with different normal pressure (7.95–17.80 kPa) and insertion velocity (1–9 mm/s), which agrees well with the experimental data. The results of the paper can help to better understand the intrinsic characteristics of the needle–tissue interactions and optimal needle actuation strategies.