This paper provides an overview of the phenomenology, longitudinal outcome data, assessment and management of separation anxiety disorder (SAD) in children and adolescents. SAD is qualitatively different from early worries, and is characterised by an abnormal reactivity to real or imagined separation from attachment figures, which significantly interferes with daily activities and developmental tasks. Different epidemiological studies indicate a prevalence of 4 to 5% in children and adolescents. In contrast to other anxiety disorders, 50 to 75% of children with SAD come from homes of low socioeconomic status. The severity of symptomatology ranges from anticipatory uneasiness to full-blown anxiety about separation, but children are usually brought to the clinician when SAD results in school refusal or somatic symptoms. School refusal is reported in about 75% of children with SAD, and SAD is reported to occur in up to 80% of children with school refusal. Longitudinal studies have suggested that childhood SAD may be a risk factor for other anxiety disorders, but whether this link is specific to, for example, panic disorder and agoraphobia, or whether SAD represents a general factor of vulnerability for a broad range of anxiety disorders is still debated. Most relevant data are reported on nonpharmacological treatments (psychoeducational, behavioural, cognitive-behavioural, family and psychodynamic), and these are the first choice approach in SAD. Controlled studies show efficacy of cognitive-behavioural therapy in children with anxiety disorders and specifically in SAD-school phobia, supporting this approach as the best proven treatment. Pharmacotherapy should be used in addition to behavioural or psychotherapeutic intervention when the child's symptoms have failed to respond to those treatments, and he/she is significantly impaired by the symptoms. Selective serotonin (5-hydroxytryptamine; 5-HT) reuptake inhibitors (SSRI) have a good adverse effect profile and may be considered as first choice drugs in SAD. When different SSRIs fail to improve symptomatology, a trial with a tricyclic antidepressant (TCA) is indicated, with careful monitoring of cardiac functioning. Because of the adverse effect profile and the potential for abuse and dependence, benzodiazepines should be used only when a rapid reduction of symptomatology is needed, until the SSRI or the TCA have begun to be effective (few weeks). Buspirone should be considered in children who have not responded to other treatments. Further research is needed to confirm efficacy of newer antidepressants (venlafaxine, mirtazapine, nefazodone) in childhood anxiety disorders.