A　critical　issue　in　performance-based　seismic　assessment　of　structures　and　infrastructures　is　the　development　of　reliable　formulations　able　to　correlate　engineering　demand　parameters　(EDPs)　with　given　earthquake　intensity　measures　(IMs).　This　task　involves　the　following　steps:　i)　selection　of　target　cases-study　and　elaboration　of　the　corresponding　structural　models,　ii)　preparation　of　the　database　collecting　seismic　records,　iii)　identification　of　candidate　EDPs　and　IMs,　iv)　nonlinear　dynamic　analyses,　v)　numerical　calibration　of　functional　models　able　to　correlate　EDPs　and　IMs,　vi)　evaluation　of　the　predictive　capability　of　the　developed　models.　Within　this　framework,　the　present　paper　exploits　an　advanced　nonlinear　regression　method　-　named　Evolutionary　Polynomial　Regression　technique　-　in　order　to　obtain　several　non-dominated　models　(according　to　the　Pareto＇s　dominance　criterion)　that　predict　EDPs　as　function　of　assigned　IMs　for　fixed-base　and　base-isolated　multi-storey　reinforced　concrete　buildings　subjected　to　ordinary　and　pulse-like　ground　motion.