Observations of a nearby star-forming region reveal that large stars are more prevalent than models have predicted. Because the most massive stars have the biggest influence on their surroundings - through ultraviolet radiation, stellar winds, supernova explosions, and production of heavy elements - this excess of large stars will have wide-ranging implications for astrophysics. Studying massive stars is difficult, because the heaviest stars have the shortest lifetimes before exploding as supernovae. Here, Fabian Schneider and colleagues analyzed a survey by the Very Large Telescope of hundreds of stars in the region 30 Doradus (30 Dor), a cluster of young stars in the Large Magellanic Cloud, a neighboring galaxy to our own Milky Way. By modeling how many stars formed at each mass, and adjusting for observational biases, the authors were able to clearly determine the fraction of massive stars that were produced. There were many more stars at high masses (>30 times the mass of the Sun) than predicted by long-standing models of star formation, and the discrepancy gets bigger at the highest masses. Such a skewed distribution, known as a top-heavy initial mass function, is expected to eventually generate many more exotic objects such as black holes and neutron stars. The authors also used the data to estimate the duration of the star-forming event, finding that the burst of activity in 30 Dor happened relatively quickly, over the course of less than 10 million years.