Two decades after the discovery that 20% of familial amyotrophic lateral sclerosis (ALS) cases were linked to mutations in the superoxide dismutase-1 ( SOD1) gene, a substantial proportion of the remainder of cases of familial ALS have now been traced to an expansion of the intronic hexanucleotide repeat sequence in C9orf72. This breakthrough provides an opportunity to re-evaluate longstanding concepts regarding the cause and natural history of ALS, coming soon after the pathological unification of ALS with frontotemporal dementia through a shared pathological signature of cytoplasmic inclusions of the ubiquitinated protein TDP-43. However, with profound clinical, prognostic, neuropathological, and now genetic heterogeneity, the concept of ALS as one disease appears increasingly untenable. This background calls for the development of a more sophisticated taxonomy, and an appreciation of ALS as the breakdown of a wider network rather than a discrete vulnerable population of specialised motor neurons. Identification of C9orf72 repeat expansions in patients without a family history of ALS challenges the traditional division between familial and sporadic disease. By contrast, the 90% of apparently sporadic cases and incomplete penetrance of several genes linked to familial cases suggest that at least some forms of ALS arise from the interplay of multiple genes, poorly understood developmental, environmental, and age-related factors, as well as stochastic events.