Gas hydrates have been known for 200 years or more and the nature of these interesting solids remained hidden until crystallographic studies some 140 years later revealed that they were crystalline guest–host materials classified as clathrates. [1] The discovery of vast quantities of natural gas hydrate in sediment offshore on the continental margins and under the permafrost has brought these materials into the public eye once again as there are implications concerning the global energy supply and the environment. Although natural gas hydrates come in three distinct structural forms, depending on the guest molecules present, herein we will be concerned with methane hydrate—by far the most extensively occurring of the three. Hydrate formation in nature requires the presence of methane, either of biogenic or thermogenic origin, and it will form hydrate under appropriate thermodynamic conditions (p, T) where there is a trapping mechanism as, for example, in sediment. So, how do natural gas hydrates form? The thermodynamic conditions under which methane hydrate can form in nature have been discussed and summarized. Under appropriate temperatures and pressures, hydrates can form at gas–liquid interfaces, as well as in methane-saturated bulk aqueous phases of characteristic salinity, perhaps mediated by solid interfaces. [2] Because of the low mutual solubility of small hydrophobic guests such as methane in water, the concentration of methane must suddenly increase by a factor of � 670 upon solid hydrate formation from a saturated solution. Due to the small size of the hydrate nuclei formed upon nucleation, gaining direct experimental information about the mechanism of hydrate formation is difficult, although some progress in this area has been made. [3] A major effort has been put into molecular dynamics simulations of hydrate formation from aqueous solutions to elucidate the mechanism and the recent work of Sum, Wu and co-workers in this area, [4] is the subject of this Highlight. From laboratory studies it is known that the initial formation of solid hydrate is delayed from the time when the thermodynamic conditions for hydrate formation are met initially. The delay period, known as the induction time, may range from seconds to days or longer. Therefore, simulations of hydrate formation tend to introduce a number of factors to shorten the induction time so that nucleation can be seen on accessible simulation time scales (usually tens of ns to a few ms). These include using higher thermodynamic driving forces for nucleation (for example, higher pressure at constant temperature), the presence of a methane gas phase adjacent to the solution, supersaturation of the water solution by methane, or the presence of curved interfaces such as at gas bubble– solution interfaces. Rodger and co-workers [5] have performed a series of molecular dynamics simulations and have recently published a comprehensive study and review of hydrate nucleation mechanisms for the methane/water system. They describe mechanisms for hydrate nucleation as being driven by either 1) water ordering around individual methane guests to form hydrate cages which then agglomerate into other cages to form the bulk hydrate phase (the labile cluster hypothesis), or