Agreement between measured response of an offshore structure and numerical predictions using an initial finite element model (IFEM) is in general poor. An algorithm is developed, which produces an updated finite element model (UFEM) that is fully correlated with respect to modal and static deflection measurements. An incremental nonlinear methodology based on large admissible perturbations in cognate space is used to produce the UFEM by postprocessing results of the initial FEA. No other FEA or trial and error are required. Iterations within each increment are used only to correct for dependence of hydrodynamic excitation on correlation variables. The UFEM corresponds to a real structure and may differ from the IFEM in response and correlation variables by 100–300 percent depending on correlation measures and structural size. Several numerical applications for three offshore structures are used to assess the strength, limitations, and cost of the large perturbation methodology.