Amyotrophic lateral sclerosis (ALS; motor neuron disease) is a relentlessly progressive disorder. After half a century of trials, only one drug with modest disease-modifying potency--riluzole--has been developed. The diagnosis of this disorder is still clinical and there is a pronounced delay between the onset of symptoms and diagnosis, possibly beyond the therapeutic window. Bedside quantification of the involvement of the corticospinal tract and extramotor areas is inadequate and functional rating scales, forced vital capacity, and patient survival have been the measures of therapeutic response so far. Potential biomarkers that are sensitive to the progression of disease, which might enhance the diagnostic algorithm and provide new drug targets, are now being identified from analysis of the blood and cerebrospinal fluid, as well as from neuroimaging and neurophysiology studies. In combination, these biomarkers might be sensitive to early therapeutic effects and would reduce our reliance on animal models, which have uncertain relevance to sporadic ALS in human beings. Such biomarkers might also resolve complexities of phenotypic heterogeneity in clinical trials. In this Review, we discuss the development of biomarkers in ALS and consider potential future directions for research.