This thesis introduces and explores a novel approach to constructing elastic multibody models using experimentally derived data, addressing limitations found in previous studies. A comprehensive review of the standard modeling approach based on linear finite element matrices and a floating frame of reference formulation sets the stage. The investigation establishes identities between the finite element method and reference frame kinematics, streamlining the derivation of the theoretical framework. The experimental modeling approach bases on modal models derived from unconstrained structures and their rigid body inertia properties. However, prior studies relying on experimentally determined modal models have encountered errors in coupling matrices between deformation and rigid body motion. To overcome this, the novel approach introduces a Buckens-type frame as the body's reference frame and extends the experimental database to include rigid body inertia properties. The experimental modeling approach is rigorously examined in three testing scenarios of increasing complexity, including a numerical study and two experimental cases involving simple flat steel and a small-scale wind turbine blade. Comparisons with conventional modeling, assuming the same number of modes, demonstrate the novel approach's effectiveness. The thesis establishes an improved experimental-based modeling approach as a compelling alternative, particularly for complex tasks where numerical models are prone to significant errors. Despite potential measurement errors directly impacting the multibody model, the approach produces more accurate models than conventional methods, especially for structurally complex systems. The study underscores the importance of accurate measurement hardware and execution. With advancements in measurement technology, experimental-based modeling emerges as a crucial simulation tool for addressing modern design complexity