Principessa Mafalda arrived in the world aware of how things can go wrong at launch. Her sister ship, Principessa Jolanda, had capsized and sunk immediately upon launching on 22 September 1907 at Cantiere Navale di Riva Trigoso. The engineers tried again. When Principessa Mafalda went down the ways on 22 October 1908, she left the yard without her superstructure installed - a cautious acknowledgment that the first launch had carried too much weight too high. She was finally completed on 30 March 1909 and named for Princess Mafalda of Savoy, second daughter of King Victor Emmanuel III. By 1910 she was the flagship of Navigazione Generale Italiana, and Guglielmo Marconi was conducting long-distance radio experiments from her decks.

For more than a decade she was the pride of the Genoa-Buenos Aires run. At eighteen knots she was genuinely fast. Her two-story ballroom and Louis XVI cabins made her desirable. During the First World War the Italian Royal Navy requisitioned her to house officers at Taranto; in 1918 she returned to South American service. By 1926 she had completed more than ninety round-trips across the Atlantic without serious incident. But the ship was aging. By 1922 she had been replaced as NGI's flagship. Maintenance was, in retrospect, no longer what it had been. The ninety-first voyage would not pass uneventfully.

On 11 October 1927 she sailed from Genoa carrying 971 passengers and 288 crew. Most of the passengers in steerage were Italian emigrants - families heading for new lives in Argentina, carrying letters, photographs, and whatever small wealth they had managed to accumulate for the passage. The ship also carried 300 tonnes of cargo, 600 bags of mail, and 250,000 gold lire destined for the Argentine government. The journey was to take fourteen days with stops at Barcelona, Dakar, Rio de Janeiro, Santos, and Montevideo. Almost immediately the ship fell behind schedule. By 23 October she had developed a visible list to port. The passengers, given few explanations, began to understand that something was very wrong beneath their feet.

At 17:15 on 25 October 1927, 80 miles off Salvador da Bahia near the Abrolhos Archipelago, the ship shuddered hard. A propeller shaft had fractured and torn open the hull. She did not sink quickly. She circled slowly for at least an hour while confused signals went out to other vessels. Rescue ships approached but kept their distance, fearing her boilers might explode. Their lifeboats came alongside to shuttle survivors to safer decks, especially the steamship Alhena. Captain Gulì directed operations from the bridge with a megaphone. The evacuation began orderly. At 22:03 the power failed, and the ship fell into darkness, and the order broke. Some lifeboats were rushed by desperate passengers and capsized. Others, poorly maintained, took on water as they touched the sea. Some passengers jumped. Sharks attacked some of them. At 22:10, four hours and twenty minutes after the first shudder, Principessa Mafalda sank stern first.

Gulì stayed on the bridge as the ship went down, fulfilling the old maritime compact in a tragedy that did not deserve his courage. He was posthumously decorated for bravery at sea, as were the two radio operators, Luigi Reschia and Francesco Boldracchi, who remained at their posts until they drowned sending distress calls. The chief engineer, Silvio Scarabicchi, reportedly shot himself. Of 1,252 souls aboard, 314 died - the largest loss of life on an Italian ship and the largest in the Southern Hemisphere in peacetime. Italian emigration history absorbed the blow the way such histories do: with photographs in family albums, with memorial masses, with the quiet weight of names never spoken again at the dinner table.

The Italian Navy Board determined that a joint in the propeller casing had failed. It ordered all Italian-registered vessels to be fitted with devices to prevent similar accidents. It noted that six stern lifeboats had been placed where they could not be launched. It did not examine the ship's age, the inadequate maintenance, or the crew's actions. NGI was ordered to compensate victims' families heavily. A 2012 analysis found a pattern the inquiry had missed: a greater proportion of first-class passengers died (51.8%) than of steerage passengers (27.8%). Survival rates for men (74.1%) and women (73.3%) were almost identical. The wreck site, somewhere on the seabed near the Abrolhos, has never been definitively confirmed.