The culprit of kava hepatotoxicity will continue to remain a mystery in humans, if the underlying reaction is of idiosyncratic, unpredictable, and dose-independent nature due potentially to some metabolic aberration in a few individuals emerging from kava use. In addition, kava hepatotoxicity is presently not reproducible experimentally in preclinical models, as demonstrated by studies showing whole kava extracts are not hepatotoxic. This led us to propose our ‘working hypothesis’ that contaminant hepatotoxins including moulds might have caused rare kava hepatotoxicity in humans. Further studies are now warranted to proof or disproof our working hypothesis, because kava hepatotoxicity possibly based on contaminant hepatotoxins could be a preventable disease. In the meantime, however, for minimizing toxicity risk in kava users, a pragmatic approach should focus on the medicinal use of an aqueous extract derived from peeled rhizomes and roots of a non-mouldy noble kava cultivar, limited to maximum 250-mg kavalactones daily for acute or intermittent use. Copyright © 2012 John Wiley & Sons, Ltd.