INTRODUCTION: This study investigated the effects of trait anxiety and state worry on working memory performance in a normal sample. Phase one investigated the effects of trait anxiety and state worry on the capacity of specific working memory components. Phase two investigated the validity of Eysenck and Calvo's (1992) Processing Efficiency Theory of worry. METHOD: Sixty adult participants (40 females and 20 males with a mean age of 26 years) were assigned to a 2 (trait anxiety: Low vs. high)x2 (state worry: Low vs. high) between-subjects design. RESULTS: Contrary to prediction, worry did not lead to a decrement in performance on verbal working memory tasks but unexpectedly enhanced performance on visual tasks in participants with low trait anxiety (LTA). The results were also in opposition to expectations for Phase two. Individuals in the conditions of high trait anxiety and/or high state worry (LTA/HW, HTA/LW, and HTA/HW) displayed shorter response latencies than individuals in the LTA and low state worry (LTA/LW) condition on both verbal and spatial working memory (i.e., N-back) tasks. CONCLUSION: Although non-pathological worry is predominantly a verbal-linguistic activity, it may also be complemented by the processing of visual imagery which facilitates problem-solving and adaptive functions.