The effects of expectancy for improvement in relaxation training were examined in a factorial experiment. Relaxation training via audio tape was compared with relaxation from a soothing music tape. These two conditions were crossed with three levels of expectancy (high, medium, and low) and the effects measured with three dependent measures of anxiety (A‐State, A‐Trait, A‐Continuum) over three 45‐minute training sessions. Results indicated significant anxiety reduction for only the relaxation groups. Although the expectancy manipulation was highly successful as indicated by a manipulation check, the participants' expectancy for improvement was not related to decrements on the anxiety measures.