Can hot water freeze faster than cold water?

Under some conditions, initially hotter water has been reported to freeze before colder water. This is called the Mpemba effect. Whether it is a genuine, reliable physical effect is still contested. Some researchers say careful experiments show nothing meaningful, while others report that specific conditions reproduce it. [1] [2]

The usual definition starts with two identical containers holding equal amounts of water, differing only in starting temperature. Both are cooled in exactly the same way. If the warmer one freezes first, the Mpemba effect has been seen. It is not claimed for every temperature, container or cooling setup: water at 99.9°C will not beat water at 0.01°C. The effect is named after Erasto Mpemba, a Tanzanian teenager who, with the physicist Denis Osborne, carried out the first systematic studies in the 1960s. Follow-up work has often failed to replicate their result. References to something similar go back at least to Aristotle, according to one paper. [1] [2] [3]

There is no universally accepted definition. Mpemba and Osborne recorded the time for freezing to start, while other studies include the whole freezing process. This inconsistency contributes to the discrepancies in the literature. [3]

A 2016 Scientific Reports paper reviewed the evidence and ran its own controlled experiments. It concluded that it found no observations that could reasonably be called the Mpemba effect, and called the effect a scientific fallacy. It argued that nearly all reported cases show only marginal differences in cooling time, and that many studies did not report the height at which temperature was measured, which could change their conclusions. It regarded the original Mpemba and Osborne dataset as possibly flawed because no one had reproduced it. [3]

The same paper noted that freezing is a large energy step, roughly equal to cooling the same water from 80°C to 0°C, and that it starts by nucleation, which is sensitive to container imperfections and impurities. It said no study had shown repeatable results when freezing is included, though individual cases of warm water freezing first have been seen. The Atlantic describes related work by Henry Burridge and Paul Linden, who found that readings depended on where the thermometer was placed. MIT's engineering site similarly says it has never been convincingly shown that hot and cold water behave differently at any step, and points to impurities such as salt, dissolved solids and gases as complicating factors. [3] [2] [4]

A 2025 study in The Journal of Physical Chemistry B reports that when cups of hot and cold water are placed together in a freezer saturated with ice-nucleating agents, the hot sample freezes faster and to a greater depth, especially when the initial temperature gap is large. Supercooling was eliminated and temperature recording was tightly controlled. The authors note that the onset of crystallization is always later for hotter samples, and they attribute the result to differences in heat transfer consistent with Newton's law. They say this confirms the original Mpemba and Osborne findings. [5]

Separately, a Nature Reviews Physics piece describes work by John Bechhoefer and colleagues showing how a Mpemba effect can be reliably produced. The search excerpt describes anecdotal, elusive reports of the effect in water, and the reliable demonstrations it refers to are for a broader, more general version of the effect, not necessarily ordinary water in a freezer. [6]

The sources agree that the result depends heavily on definitions, measurement details and conditions such as nucleation and impurities. They disagree on whether any robust effect remains once these are controlled. The search results do not show a settled consensus, so the effect is best described as unresolved for ordinary water. [2] [3] [5]