The 1755 Lisbon earthquake is still poorly understood due to its offshore location and complex macroseismic intensity pattern. Gutenberg and Richter (1949) tentatively assigned a magnitude between 8¾ and 9 judging from an estimated perceptibility radius of 2500 km. More recent attempts to estimate the magnitude from isoseismal areas led to results in the 8.5–8.7 range. These values have been adopted in several studies of the seismic hazard of southwest Iberia. In this article, I use the available macroseismic data—a total of 1206 data points from Portugal, Spain, and Morocco—to reassess the magnitude of the 1755 Lisbon earthquake. Because a training set of instrumental earthquakes was not available, I apply the technique of Bakun and Wentworth (1997) in conjunction with the ground‐motion model of Atkinson and Wald (2007), which was selected through comparison with the 1969 M 7.8 Gorringe bank earthquake data. I obtain a moment magnitude of 7.7±0.5⁠, significantly lower than previous results. The epicentral location obtained with this analysis is offshore southwest Iberia—as expected given the tsunami effects—but poorly constrained. Based on the pattern of intensity data outliers, I suggest that the source was complex and spatially distributed, with part of the rupture taking place onshore or inshore. I propose an explanation for the large tsunamigenic power of the earthquake, which invokes the basal rupture of the Gulf of Cadiz accretionary prism. Finally, I discuss the implications for hazard assessment of the type of complex rupture proposed.