In the late summer of 1918, the devastating second wave of the Spanish flu arrived on America’s shores. Carried by World War I doughboys returning home from Europe, the newly virulent virus spread first from Boston to New York and Philadelphia before traveling West to infect panicked populations from St. Louis to San Francisco.

Lacking a vaccine or even a known cause of the outbreak, mayors and city health officials were left to improvise. Should they close schools and ban all public gatherings? Should they require every citizen to wear a gauze face mask? Or would shutting down important financial centers in wartime be unpatriotic?

When it was all over, the Spanish flu killed an estimated 675,000 Americans among a staggering 20 to 50 million people worldwide. Certain U.S. cities fared far worse than others, though, and looking back more than a century later there’s evidence that the earliest and most well-organized responses slowed the spread of the disease—at least temporarily—while cities that dragged their feet or let down their guard paid a heavier price.

By mid-September, the Spanish flu was spreading like wildfire through army and naval installations in Philadelphia, but Wilmer Krusen, Philadelphia’s public health director, assured the public that the stricken soldiers were only suffering from the old-fashioned seasonal flu and it would be contained before infecting the civilian population.

When the first few civilian cases were reported on September 21, local physicians worried that this could be the start of an epidemic, but Krusen and his medical board said Philadelphians could lower their risk of catching the flu by staying warm, keeping their feet dry and their “bowels open,” writes John M. Barry in The Great Influenza: The Story of the Deadliest Pandemic in History.

As civilian infection rates climbed day by day, Krusen refused to cancel the upcoming Liberty Loan parade scheduled for September 28. Barry writes that infectious disease experts warned Krusen that the parade, which was expected to attract several hundred thousand Philadelphians, would be “a ready-made inflammable mass for a conflagration.”

Krusen insisted that the parade must go on, since it would raise millions of dollars in war bonds, and he played down the danger of spreading the disease. On September 28, a patriotic procession of soldiers, Boy Scouts, marching bands and local dignitaries stretched two miles through downtown Philadelphia with sidewalks packed with spectators.

Just 72 hours after the parade, all 31 of Philadelphia’s hospitals were full and 2,600 people were dead by the end of the week.

George Dehner, author of Global Flu and You: A History of Influenza, says that while Krusen’s decision to hold the parade was absolutely a “bad idea,” Philadelphia’s infection rate was already accelerating by late September.

“The Liberty Loan parade probably threw gasoline on the fire,” says Dehner, “but it was already cooking along pretty well.”

The public health response in St. Louis couldn’t have been more different. Even before the first case of Spanish flu had been reported in the city, health commissioner Dr. Max Starkloff had local physicians on high alert and wrote an editorial in the St. Louis Post-Dispatch about the importance of avoiding crowds.

When a flu outbreak at a nearby military barracks first spread into the St. Louis civilian population, Starkloff wasted no time closing the schools, shuttering movie theaters and pool halls, and banning all public gatherings. There was pushback from business owners, but Starkloff and the mayor held their ground. When infections swelled as expected, thousands of sick residents were treated at home by a network of volunteer nurses.

Dehner says that because of these precautions, St. Louis public health officials were able to “flatten the curve” and keep the flu epidemic from exploding overnight as it did in Philadelphia.

“It’s that crush of new cases in such a short period of time that completely overwhelms a city’s capacity,” says Dehner. “That magnifies whatever problems you’re already having.”

According to a 2007 analysis of Spanish flu death records, the peak mortality rate in St. Louis was only one-eighth of Philadelphia’s death rate at its worst. That’s not to say that St. Louis survived the epidemic unharmed. Dehner says the midwestern city was hit particularly hard by the third wave of the Spanish flu which returned in the late winter and spring of 1919.

In San Francisco, health officials put their full faith behind gauze masks. California governor William Stephens declared that it was the “patriotic duty of every American citizen” to wear a mask and San Francisco eventually made it the law. Citizens caught in public without a mask or wearing it improperly were arrested, charged with “disturbing the peace” and fined $5.

In his book, Barry says that the gauze masks city officials claimed were “99 percent proof against influenza” were in reality hardly effective at all. San Francisco’s relatively low infection rates in October were probably due to well-organized campaigns to quarantine all naval installations before the flu arrived, plus early efforts to close schools, ban social gatherings and close all places of “public amusement.”

On November 21, a whistle blast signaled that San Franciscans could finally take off their masks and the San Francisco Chronicle described “sidewalks and runnels… strewn with the relics of a tortuous month.”

But San Francisco’s luck ran out when the third wave of the Spanish flu struck in January 1919. Believing masks were what saved them the first time, businesses and theater owners fought back against public gathering orders. As a result, San Francisco ended up suffering some of the highest death rates from Spanish flu nationwide. The 2007 analysis found that if San Francisco had kept all of its anti-flu protections in place through the spring of 1919, it could have reduced deaths by 90 percent.