Although freelisting and semi-structured interviews are widespread methods in ethnobotany, few studies quantitatively examine how these methods may bias results. Using a comprehensive ethnobotanical inventory of palm species, uses and names in the Chácobo tribe of Bolivia, we show that interviews elicit more items than freelists, but the effect is sensitive to sample size, item type and data categorization. This implies that even subtle methodological choices may greatly affect reported results.