ACL has a fundamental role in knee biomechanics and ACL insufficiency is probably the most common knee pathology. Nevertheless complete knowledge of ACL biomechanics is not achieved and predictions on ACL behaviour or optimal treatment are hard and still controversial, although of primary importance for a safe and good individual reconstruction. In this paper we describe the construction and validation of a new ACL model, computable by its anatomical description and aiming at reliable predictions in a virtual surgical environment. In our model, ACL is made of 9 curvilinear connected fibers, corresponding to the individual surface fibers, which are described as a succession of 20 punctual physical particles linked by linear viscoelastic relations, and they can twist, bend and stretch under applied forces. The model has been validated on 4 fresh pig knees and has given very consistent results to describe the ligament's anatomy, function and mechanics. Simulations of passive kinematics give interesting results on fibers elongations, forces and ligament deformations, confirming the expected correlation between fibers' strain and stress. The advantage and the feature of this method with respect to previous models is the possibility to take into account more accurately ACL anatomy, mechanical properties and the ligament behaviour during the whole range of motion. All input are geometrical data, that can be acquired in fixed position. The output is the simulation of ACL that describes the forces exerted and respects sizes and shapes of the ligament. The model accuracy and the simple input it requires let us envisage a useful application on human ACL to develop surgical planning and simulation.