Early humans evolved to throw about two millions years ago, according to new research.
Anatomy changes found in the extinct species Homo erectus allowed this ability to develop.
Archaeological evidence suggests hunting intensified during this time, which scientists now attribute to the ability to throw.
Researchers tell the journal Nature that throwing helped early hunters to evolve and migrate around the globe.
The ability to throw at very high speeds is unique to humans. We can throw much faster than our closest living relative - the chimpanzee - which can only reach speeds of 20mph compared to 90mph that many professional athletes can reach.
To investigate the evolutionary development of the ability to throw, scientists first had to understand the biomechanics of throwing today.
They recorded the throwing movements of college baseball players using motion capture cameras and observed that the shoulder acts like a slingshot as the arm rotates backwards.
The ligaments and tendons surrounding the shoulder then stretch and store elastic energy, which powers the forward throw. When this energy is released it generates what scientists found was the fastest motion the human body produces.
Neil Roach, from George Washington University, US, who led the study, said that changes in the anatomy of hominins (early humans) that occurred two million years ago, enabled energy storage in the shoulder that allowed fast throwing, and therefore hunting, to occur.
"Success at hunting allowed our ancestors to become part-time carnivores, eating more calorie-rich meat and fat and dramatically improving the quality of their diet.
"This dietary change led to seismic shifts in our ancestors' biology, allowing them to grow larger bodies, larger brains, and to have more children, and it also did interesting things to our social structure.
"We start to see the origins of divisions of labour around that time, where some would be hunting, others would be gathering new foods.
"It probably also allowed us to move to new environments, such as areas that did not have vegetation to support us before we had the ability to hunt," Dr Roach told BBC News.
He added that it was important to remember "that what we think about hunting and behaviour is still a hypothesis" and further studies were needed.
Another member of the research team, Daniel Lieberman from Harvard University, US, said the most fascinating finding for him was that half the power that humans generate from throwing comes from elastic energy stored in the shoulder.
"That's not a by-product of evolution for something else, it's clearly an adaptation. There were shifts in our anatomy that enabled us to throw accurately, so we want to understand better just what those early hunting challenges were."
Speaking to BBC News from Kenya he explained how, compared to the wildlife he was surrounded by - cheetahs, lions and leopards - humans had no natural weapons such as claws.
"Human hunting is such an fascinating problem and the fact that these features all appear by the time Homo erectus evolved, suggests that hunting may have been a selective force for the ability for throwing."
Prof Lieberman added that the next step was to discover what exactly early humans were using to hunt, as no weapons have been found from this time in the archaeological record.
Susan Larson from Stony Brook University, New York, was not involved in the research. Her work also focuses on the shoulder anatomies of primates and humans in order to study their evolution. She said it could be easy to over interpret what was significant from a fossil.
"We're looking at the same fossils, it's a question about how you interpret the anatomy that you see in those fossils.
"Homo erectus was not necessarily such a proficient thrower. I think [Dr Roach and colleagues] are discounting the combination of features of how the shoulder as a whole has to work.
"Their [Homo erectus] shoulder still functioned to give the hand a very wide range of motion in order to manipulate things and make tools, but it wasn't designed the same way as it is in humans.
"You cannot look at just one thing and say how a complicated piece of anatomy works, especially something like the shoulder. You have to understand how all parts work together to bring about a broad range of motion," Prof Larson told BBC News.
The new research introduces more of a strong theoretical model, than any definitive evidence of throwing behaviour in Homo erectus, said Jill Rhodes from Drew University, US.
"The humeral torsion - the angle at which the head of the humerus (top of the arm) is articulating at the shoulder joint - does not fall within the range of modern throwing athletes and more relevantly, as throwing is a one-handed behaviour, there is no evidence for asymmetry in the humeral torsion angle in Homo erectus.
"There simply is not the fossil preservation and without it, we cannot say that the humeral torsion demonstrated in that species is not an aspect of phylogeny rather than behaviour.
"This research opens a window into our understanding of past behaviour, but our view is still cloudy."