JupiteX Get the app
Health & Medicine18 Nov 2026 · about 7 min

World of Warcraft's virtual 'Corrupted Blood' pandemic became a model for real epidemics

The brief

In September 2005, Corrupted Blood escaped from a World of Warcraft dungeon and spread through the game’s virtual cities. The spell was designed for a single fight against the boss Hakkar, not to become a citywide outbreak. It drained characters’ health and passed to nearby characters. Thousands of characters died, turning a programming error into an unexpected epidemic. The bug involved pets belonging to hunters and warlocks. A player could dismiss an infected pet and later summon it in a capital city. The disease returned with the pet, still able to infect nearby characters. Inside the raid dungeon, Corrupted Blood was an obstacle players could manage; beyond it, the effect was never meant to exist. Pets therefore became a route for the disease to escape its intended setting. Blizzard acknowledged the problem on 16 September and tried barriers around affected areas, but the disease was too contagious to contain that way. After several server resets and a patch making pets immune, the outbreak was declared eradicated on 8 October. Studies published in 2007 examined what the accidental outbreak could reveal about disease spread and human behaviour.

01

What was the Corrupted Blood outbreak, and what bug let the disease spread beyond its dungeon?

In September 2005, Corrupted Blood escaped from a World of Warcraft dungeon and spread through the game’s virtual cities. The spell was designed for a single fight against the boss Hakkar, not to become a citywide outbreak. It drained characters’ health and passed to nearby characters. Thousands of characters died, turning a programming error into an unexpected epidemic.

The bug involved pets belonging to hunters and warlocks. A player could dismiss an infected pet and later summon it in a capital city. The disease returned with the pet, still able to infect nearby characters. Inside the raid dungeon, Corrupted Blood was an obstacle players could manage; beyond it, the effect was never meant to exist. Pets therefore became a route for the disease to escape its intended setting.

Blizzard acknowledged the problem on 16 September and tried barriers around affected areas, but the disease was too contagious to contain that way. After several server resets and a patch making pets immune, the outbreak was declared eradicated on 8 October. Studies published in 2007 examined what the accidental outbreak could reveal about disease spread and human behaviour.

02

How large was the outbreak, and how many people were playing World of Warcraft at the time?

The outbreak killed thousands of game characters, but the article does not give a precise total for deaths or infections. That distinction matters: the scale of the epidemic is described through its widespread impact, not a complete count of cases. The disease reached virtual cities, where it could spread among nearby characters.

At the time, World of Warcraft had more than four million subscribers. That figure gives a sense of the game’s large population, but it is not a count of players infected or characters killed. The article reports that Blizzard acknowledged the problem on 16 September, after Corrupted Blood had escaped the Zul’Gurub raid dungeon.

Epidemiologists later studied the event, but the spread had not been deliberately tracked. A commonly repeated account says the US Centers for Disease Control and Prevention asked Blizzard for data, believing the outbreak was a planned simulation; the article notes that no data could be provided. The 2007 studies therefore examined an accidental event whose full spread had not been recorded.

03

What happened to vulnerable characters, and how did players and Blizzard try to respond?

Corrupted Blood affected characters differently according to their level. For a high-level character, the disease was little more than a cold. For a low-level character, it could mean death in a couple of seconds. The article uses this contrast to show that the most vulnerable characters suffered most, much as vulnerable people do in epidemics.

Blizzard first tried quarantine barriers around the areas most affected. The approach failed: the disease was too contagious, the map could not be truly sealed off, and players were not willing to cooperate. Players also carried the effect between cities as they travelled, making it difficult to isolate the outbreak within one place.

Blizzard eventually used tools unavailable to a real health authority: it restarted the world. After several complete server resets and a patch that made pets immune, the outbreak was declared eradicated on 8 October. The episode showed both the limits of quarantine in a virtual world and the unusual ability of a game operator to reset that world. Epidemiologists later studied the event, though its spread had not been recorded.

04

How did infected pets, symptomless NPCs, crowded cities, and instant travel help the disease spread?

Several game mechanics combined to spread Corrupted Blood. Hunters and warlocks could dismiss infected pets and summon them again in a capital city, where the effect returned. The article compares this route to a zoonosis, with pets acting as a reservoir. Meanwhile, non-player characters such as guards and vendors could be infected without becoming ill or dying.

Those NPCs were symptomless carriers in busy locations. Players travelling between cities could also carry the disease, and travel was instant. In packed capital cities, transmission depended on crowd density, just as it would in a real city. These factors brought infected characters, susceptible characters, and carriers together in places with many people.

Eric Lofgren and Nina Fefferman calculated that the reproduction rate in those hubs reached 100 infections per hour. They considered that rate impossible for a real pathogen. The outbreak ended only after several complete server resets and a patch that made pets immune. Its unusual combination of pets, carriers, crowds, and easy transport helped make it a subject of epidemiological study.

05

Why did epidemiologists compare the virtual outbreak to diseases such as bird flu and SARS?

Epidemiologists compared the virtual outbreak with real diseases because it offered examples of transmission through animal carriers and travel. In March 2007, Ran Balicer pointed to similarities with bird flu: birds such as ducks can carry the virus without becoming ill. In the game, infected pets helped carry Corrupted Blood beyond its intended dungeon.

Balicer also compared the outbreak with SARS, which spread from one country to another between 2002 and 2004 aboard aircraft. In World of Warcraft, players could travel instantly between cities, including while infected. The game’s crowded capital hubs added another route for transmission, as many characters were close together.

These comparisons did not make the virtual disease identical to a real pathogen. The game produced features useful for examining spread, but the article notes that the calculated rate of 100 infections per hour was impossible for a real pathogen. The 2007 studies treated the accidental outbreak as a way to observe behaviour and transmission that ordinary disease models cannot fully capture.

06

What is a zoonosis, and what is an asymptomatic carrier?

A zoonosis is a pathogen that jumps from animals to people and back again. The article uses this term for the route Corrupted Blood took through hunters’ and warlocks’ pets. The pets could carry the effect out of the dungeon and into a capital city, where it could infect characters nearby.

An asymptomatic carrier is infected but does not show signs of illness. In the game, non-player characters such as guards and vendors could become infected without falling ill. They also could not die, so they remained in busy places and could pass on the disease. The article describes them as symptomless carriers; a guard in Ironforge was nicknamed its “Typhoid Mary.”

The two concepts describe different parts of the outbreak’s spread: pets provided a route between settings, while infected NPCs could remain present without obvious symptoms. Together, they made Corrupted Blood resemble features of real epidemics. The 2007 studies examined these parallels, although the virtual outbreak was accidental and its spread had not been recorded.

07

Why can’t disease-spread models alone predict how people will behave during an epidemic?

Disease-spread models can calculate how a pathogen passes from one person to another. But the article says they do not predict what each individual will do, especially when people know they might be infected. Behaviour matters because people may travel, gather, or respond in ways that change how an outbreak spreads.

Corrupted Blood gave epidemiologists a rare, accidental view of those choices. Infected players travelled between cities, and players did not cooperate with Blizzard’s quarantine barriers. The outbreak also spread through pets and infected non-player characters. These game events exposed human behaviour alongside transmission, without anyone having deliberately planned the outbreak as a simulation.

The article says releasing a real pathogen just to observe people is unthinkable, while computer models use virtual populations. World of Warcraft offered live behaviour, but it had a major limitation: nobody recorded the outbreak’s spread, so there was no data for Blizzard to provide. Studies published in 2007 still examined the event, showing its value as an unusual case, not a complete replacement for models or real-world evidence.

This brief was written by AI from the original reporting and checked by other models. Names, figures and quotes come from the source; read it for full context.

Read more in the JupiteX app

Pulse is free. New stories every 4 hours, each one broken into the questions that explain it.

Or read more news on the web