AbstractTimber floors are prone to exhibit vibration levels, which can cause discomfort to the occupants. In the last 20 years, ambient vibration tests have become very popular due to the many advantages they have over traditional forced vibration tests when dealing with civil engineering structures. Furthermore, sensitivity analyses and black-box” optimization algorithms can support the development of refined finite-element models that accurately predict the structures’ responses based on the experimental modal parameters. However, applications of these methods and techniques to timber structures are scarce compared with traditional materials. This paper presents and discusses the findings of an experimental testing campaign on a lightweight timber floor. At first, each component of the assembly was tested separately under different boundary conditions. Then, the authors evaluated the behavior of the whole floor assembly. In a second step, the authors carried out a covariance-based sensitivity analysis of finite element (FE) models representative of the tested structures by varying the different members’ mechanical properties. The results of the sensitivity analysis highlighted the most influential parameters and supported the comparison among diverse FE models. As expected, the longitudinal modulus of elasticity is the most critical parameter, although the results are very dependent on the boundary conditions. Then automatic modal updating algorithms tuned the numerical model to test results. As a concluding remark, the experimental and numerical results were compared with the outcomes of a simplified analytical approach for the floor’s first natural frequency estimate based on current European standards.