Background: Inclusion of obsessive-compulsive disorder (OCD) as an anxiety disorder in DSM-IV assumes that anxiety is the primary symptom of OCD; however, persuasive empirical evidence in support of this view has not been presented yet. In the present study we hypothesized that provoked anxiety symptoms respond better to intravenous diazepam than would provoked obsessions. We, therefore, reasoned that anxiety symptoms are secondary symptoms of OCD. Methods: To test the hypothesis we designed a double-blind, randomized, placebo-controlled crossover study. Patients underwent four experimental conditions in which the sequence of symptom provocation and IV injection of (placebo or diazepam) were alternated. Baseline and IV injection-induced symptom changes were assessed using visual analogs. Results: Obsessions and anxiety correlated strongly for all four experimental conditions in which the sequence of the symptom provocation and diazepam IV injections was alternated. IV diazepam injection before and after symptom provocation failed to preferentially modulate anxiety symptoms over obsessions. Unexpectedly, in the group in which IV diazepam injection preceded the symptom provocation, reduction of mean obsessions was even more pronounced. Conclusions: Strong correlations between anxiety and obsessions at baseline, during symptom provocation, and after IV diazepam infusion suggest that anxiety and obsessions are tightly coupled phenomena in OCD.