In May 2012, two earthquakes (M_w 6.1 and 5.9) affected the Po Plain, Italy. The strongest shock produced extensive secondary effects associated with liquefaction phenomena. Few weeks after the earthquakes, an exploratory trench was excavated across a levee of the palaeo-Reno reach, where a system of aligned ground ruptures was observed. The investigated site well preserves the geomorphic expression of a fluvial body that mainly formed in the fifteenth to sixteenth centuries as historical sources and radiometric data testify. In the trench several features pinpointed the occurrence of past liquefaction events: (i) dikes filled with overpressured injected sand and associated with vertical displacements have no correspondence with the fractures mapped at the surface; (ii) thick dikes are buried by the plowed level or even by fluvial deposits; (iii) although some of the 2012 ground fractures characterized by vertical displacement and opening occurred in correspondence of thick dikes observed in the trench, sand and water ejection did not occur; (iv) some seismites (load casts) were observed in the trench well above the 2012 water level. The results strongly suggest that shaking has locally occurred in the past producing a sufficient ground motion capable of triggering liquefaction phenomena prior to, and likely stronger than, the May 2012 earthquake. Historical seismicity documents three seismic events that might have been able to generate liquefaction in the broader investigated area. Based on the analysis of their macroseismic fields, the 17 November 1570 Ferrara earthquake is the most likely causative event of the observed palaeoliquefactions.