The past fifteen years have led to a profound re-consideration of the molecular and cellular basis of amyloid diseases. Since the formulation of the amyloid hypothesis in 1991-1992, increasing interest was initially focused at amyloid fibrils and, subsequently, at their precursors, oligomers and pre-fibrillar aggregates as main culprits of cell impairment and demise, particularly in neurodegenerative diseases with amyloid deposition. In 2002, this concept was generalized by the demonstration that pre-fibrillar aggregates were toxic even when they were grown from proteins not associated with amyloid disease. Presently, the general structural features and polymorphism of amyloid fibrils grown from a range of different peptides and proteins are rather well known; however, in spite of the growing interest in amyloid oligomers as the main source of amyloid toxicity, a better definition of their structural features remains elusive due to their transient nature, remarkable instability, high flexibility and structural heterogeneity possibly resulting in the appearance of polymorphic assemblies. Nevertheless, recent studies have started to unravel this key topic by providing significant insights into some general structural features and conformational polymorphism of amyloid oligomers and the higher order structures they generate. Important clues into the structure-toxicity relation of amyloids, the role performed by natural surfaces in oligomer growth and the molecular basis of oligomer-membrane interaction are also emerging.