Stroke is common, with a high rate of disability and death, and current therapies are both highly time sensitive and carry some risk, making rapid diagnosis important. Many cases of stroke are difficult to diagnose clinically during the acute phase, and there are limitations to the ability of diagnostic imaging to help with the necessary speed. A reliable and valid biomarker would be invaluable for this common clinical situation, as it has been with myocardial infarction. A large number of molecules have been evaluated for this role, both in the laboratory and in patients, but the results to date have been disappointing. In this article, we review the operative concepts in the search for a diagnostic biomarker of stroke, a few of the promising candidates and the current challenges to validation.