The present study evaluated the main and interactive effects of level of smoking (cigarettes per day) and anxiety sensitivity (fear of anxiety and anxiety related sensations) in predicting panic and anxiety variables in an epidemiologically-defined sample of smokers from Moscow (n=95). The combination of high levels of anxiety sensitivity and smoking predicted agoraphobic avoidance, but not frequency of panic attacks during the past week. These findings suggest anxiety sensitivity may moderate the relation between level of smoking and prototypical panic psychopathology variables (panic attacks and agoraphobic avoidance) even after controlling for the theoretically-relevant factors of alcohol abuse and negative affect.