The reanalyses of all the primary historical sources of the catastrophic March--June 1638 Calabrian earthquakes permitted to split up the seismic sequence in different events that occurred in conterminous areas and to direct our geological surveys of the June shock inside the Sila massif (northern Calabria). We carried out paleoseismological analyses along the Lakes fault (LF), a previously unknown fault. LF (trending NW-SE) cuts the eastern sector of the Sila massif and dips southwestward, damming the drainage network that runs toward the Ionian Sea. We opened four trenches close to or inside some of the small marshes and ponds created by the footwall uplift, finding evidence of several displacement events, the last one being the June 1638 earthquake (M = 6.7). The fault location fits with the historical description of a long fracture opened during the event, our trenching area being still named by the natives the earthquake sag. The young geomorphic expression of LF recalls that of the 1980 Irpinia fault, suggesting its relatively recent activation within the extensional regime characterizing the Calabrian arc. Conversely, the northern continuation of LF (Cecita Lake fault), which we argue to be active too, is not associated to any historical earthquake of the Italian seismic catalogue, this fact evidencing a long, at least 1000 years, elapsed time. Our study permits (1) the relocation of the June 1638 earthquake inside the Sila massif, the epicenter being shifted with respect to the macroseismically derived one, (2) the revision of the whole 1638 sequence, (3) the discovery of the LF, (4) the recognition of the evidence of the June 1638 surface faulting plus four previous paleoearthquakes, and (5) the evaluation of its slip rates and return interval. Finally, the relationship of LF with the neighboring faults casts light to the seismotectonic of the region, providing a new insight to the seismic hazard analysis.