We present a case history of the complicated strong-motion responses at a site with an almost 1D structure (Buia) in the near field of four Mw 5.2–6.0 earthquakes in the Friuli Plain of Italy. In one case (15 September 1976, 09:21 UTC, Mw 6.0), the 47-m alluvial cover of Buia experienced spectral amplitudes lower than its reference site on rock due to the different back azimuths from the source. The largest peaks at Buia occurred during an Mw 5.6 event. For practical purposes, we tested (1) subjectivity when preparing the 1D models (using a blind test with three experts); (2) 1D linear, linear-equivalent, or nonlinear modeling in common practice; and (3) the current seismic regulations. We also used noise recordings and the weak-motion velocities recorded for local events (with digital triplets on the top and the bottom of the alluvial cover). Buia’s response to weak events was more stable than its response to strong ones. A simple 1D model (five-layer) under linear conditions with good knowledge of the propagation velocities was able to reproduce the frequencies of the two relative strong-motion maxima with an average amplitude underestimation of 1.5. The nonlinear approach simulated the two recorded maxima with slight shifts in frequency. Because the disaggregation of the seismic hazard showed that the four destructive shocks studied represent the magnitude and distance ranges that dominate the hazard of Buia, we can conclude that the response spectra of the new Italian (and European) rules were sufficiently preventive, notwithstanding the aforementioned complications in the near field.