The aim of this project was to investigate whether structures designed for earthquake loads were robust too. Two steel structures with ten stories and five bays in each direction and with eccentric and concentric braces in the facades respectively have been analyzed using pushover analyzes. The performance for a given ground acceleration have been found using a response spectrum reduced due to the ductile behavior found in the pushover analysis with the method from ATC-40. The robustness has been assessed using a deterministic, probabilistic and risk based approach, where intact and damaged structures were analyzed. In the deterministic assessment the robustness was analyzed in relation to static and seismic horizontal loads, and the influence of ductility and hardening of the material was investigated. In the probabilistic assessment aleatoric and epistemic uncertainties were taken into account in the estimation of the annual probability of failure due to earthquakes, and two different robustness indices were calculated, one based on the probability of failure and one based on the reliability index. Finally a risk based approach was used where both direct and indirect consequences were taken into account. The analyzes have shown that the damaged structures will often have a behavior that is different from that of the original structure, because the yielding mechanisms does not work the way they are designed to when members are missing. Thus the global ductile behavior for damaged structures was found to have great importance for the earthquake related robustness, and the structural configuration was especially important for the performance. This means that structures designed for seismic actions will not necessarily be robust towards seismic loads, as this depends on the structural configuration.