The recent decision by the French government to compensate 3 recipients of hepatitis B vaccine preceding the onset of multiple sclerosis presumes a possible causal link and brings into question the use of current rules of causality assessment. Available evidence does not support a causal link or is equivocal but the accuracy of current methods of vaccine surveillance should be urgently improved. Larger and longer randomised trials, updated summaries of evidence, linked databases, prospective vaccination registers, bar-coding of vaccines and standardisation of adverse event definitions are possible measures to address current problems.