The background to the concept of the term "molten globule" as a description of intermediates observed in the folding of globular proteins is discussed. These compact intermediates are characterised by certain properties including the presence of secondary structure and considerable conformational mobility compared to the native, functional state. Those intermediates that are thermodynamically stable under mild denaturing conditions have many features in common with the transient intermediates that accumulate significantly during the process of folding. Attention is drawn to cases where the two types are however distinguished on grounds of their Stokes radius, in which cases there is currently no direct evidence for the involvement of the stable intermediates on the folding pathway. Experimental evidence relating to the early stages in folding is reviewed and compared, highlighting the temporal relationship between general collapse of the polypeptide chain and the formation of secondary structure. The continued use of the term "molten globule" is recommended where the minimum essential structural criteria for this state are met.