The damage suffered by steel structures during the Northridge (1994) and Kobe (1995) earthquakes indicates that the fully restrained (FR) connections in steel frames did not behave as expected. Consequently, researchers began studying other possibilities, including making the connections more flexible, to reduce the risk of damage from seismic loading. Recent experimental and analytical investigations pointed out that the seismic response of steel frames with partially restrained (PR) connections might be superior to that of similar frames with FR connections since the energy dissipation at PR connections could be significant. This beneficial effect has not yet been fully quantified analytically. Thus, the dissipation of energy at PR connections needs to be considered in analytical evaluations, in addition to the dissipation of energy due to viscous damping and at plastic hinges (if they form). An algorithm is developed and verified by the authors to estimate the nonlinear time-domain dynamic response of steel frames with PR connections. The verified algorithm is then used to quantify the major sources of energy dissipation and their effect on the overall structural response in terms of the maximum base shear and the maximum top displacement. The results indicate that the dissipation of energy at PR connections is comparable to that dissipated by viscous damping and at plastic hinges. In general, the maximum total base shear significantly increases with an increase in the connection stiffness. On the other hand, the maximum top lateral displacement $U_{max}$ does not always increase as the connection stiffness decreases. Energy dissipation is considerably influenced by the stiffness of a connection, defined in terms of the T ratio, i.e., the ratio of the moment the connection would have to carry according to beam line theory (Disque 1964) and the fixed end moment of the girder. A connection with a T ratio of at least 0.9 is considered to be fully restrained. The energy dissipation behavior may be quite different for a frame with FR connections with a T ratio of 1.0 compared to when the T ratio is 0.9. Thus, for nonlinear seismic analysis, a T ratio of at least 0.9 should not be considered to be an FR connection. The study quantitatively confirms the general observations made in experimental results for frames with PR connections. Proper consideration of the PR connection stiffness and other dynamic properties are essential to predict dynamic behavior, no matter how difficult the analysis procedure becomes. Any simplified approach may need to be calibrated using this type of detailed analytical study.