The role of genetic factors in anxiety disorders is yet unknown, although their is a clear familial aggregation, and although twin studies found genetic factors to be at least as important as familial factors for some anxiety disorders such as panic disorder or phobias. In this review, the complexity of genetic studies on anxiety disorders are presented. The problem of incomplete penetrance (only a fraction of cases carrying a given gene manifest a specified phenotype) is illustrated by familial studies on some stress disorder, such as post-traumatic stress disorder. The presence of a severe stress is necessary to be able to manifest the disorder, familial studies on PTSD should thus use analysis with censored data. Questions concerning phenotype specificities are also raised. For exemple, on the basis of twin samples, generalised anxiety disorder and separation anxiety disorder seem to share most of their genetic determinism. Furthermore, more major depressive disorders are found in families of probands with generalised anxiety disorder, suggesting commun vulnerability factors for some of anxiety and mood disorders. Lastly, as standardised diagnostic criteria are much more based on clinical evidence than on etiopathogenic validity, genetic factors may be involved in various bounderies of the phenotype, for example taking into account level of severity, age at onset, or comorbidity. Some of the association studies in humans, analysing the impact of genetic polymorphisms on the existence of anxiety disorders, were either negative or showed week evidence for commun traits such as nevroticism. For animal studies, quantitative trait loci technics may help to pinpoint some still unknown candidate genes. In conclusion, the genetics of anxiety disorders reflect all the difficulties proper to the genetics of complex disorders. Progress in molecular genetics will nevertheless probably change the diagnostic criteria, which are more based on clinical evidence rather than on etiological originality.