Objectives. In aromatherapy, lavender aroma is reputed to assist with relaxation. However, while there is much anecdotal evidence to that effect, the empirical literature is very inconsistent. Failure to employ adequate placebos, proper blinding, objective measures, or screening of prior beliefs about aromatherapy means that many previous findings could have been influenced by expectancy biases. The present study sought to establish whether lavender aroma and/or expectancies affect post-stress relaxation.Design. A double-blind, 3 (aroma) x 3 (instruction) x 10 (time in minutes) mixed-factorial placebo-controlled trial.Method. In a laboratory, 96 healthy undergraduate women were exposed to lavender, placebo, or no aroma during physiologically assessed relaxation after an arousing cognitive task. Where an aroma was presented, an instructional priming procedure was used to manipulate participants' expectancies about the aroma's likely impact on their ability to relax.Results. Results showed no effect of aroma on galvanic skin response during relaxation. However, the nature of instructional prime was associated with relaxation patterns: when expecting the aroma to inhibit them, participants relaxed more; when expecting facilitation, participants relaxed less. The effect was not seen with regard to self-reported relaxation (as represented by changes in state anxiety) and was independent of ratings of attitudes towards aromatherapy.Conclusions. The findings imply that the previous associations of lavender aroma with assisted relaxation may have been influenced by expectancy biases, and that the relevant expectancies are easily manipulable.