Careful clinical delineation and advances in analytical methods have opened new possibilities for the detection of inherited neurometabolic disorders, some of which require specific CSF analyses for diagnosis. Although patients suffering from these disorders have recognizable phenotypes, there are strong indications that remain many undiagnosed, leading to a continuation of futile diagnostic searches and, for most disorders, withholding of available rational therapy. As there is still widespread uncertainty about when to perform specialist CSF investigations, it is the aim of this paper to define the place for CSF investigations in the diagnostic work-up of a child with an encephalopathy of unknown origin. Most neurometabolic disorders can be identified through serum, plasma and urine analyses in conjunction with neuroradiological investigations. Whenever CSF investigations are performed, the analysis should include quantitative determination of lactate, pyruvate and amino acids, the latter by methods especially suited for CSF, in addition to cells, glucose, protein, immunoglobulin classes, specific immunoglobulins, and an evaluation of the blood-brain barrier. If the disease course is non-progressive or if extracerebral symptoms are present in addition to an encephalopathy, e.g. endocrinological, hepatic, muscular or renal symptoms, investigations of metabolites in CSF over and above lactate, pyruvate and amino acids are generally noncontributary. Specific CSF investigations, which are discussed in detail, test metabolic pathways of brain metabolism, especially of neurotransmission. For a successful diagnosis of these defects, analyses must be planned individually, before CSF samples are taken, based on family history, clinical findings and disease course. Different determinations require different logistics from taking of the sample to shipment. One indication for specialized CSF analyses including biogenic monoamines and GABA is severe neonatal/infantile epileptic encephalopathy. In addition to a therapeutic trial of B6, folinic acid should be tried empirically for two to three days as the emerging syndrome of folinic acid responsive seizures appears to be the underlying cause in a sizable proportion of patients. In later infancy and childhood, defects in the metabolism of the biogenic monoamines may be suspected in patients with (fluctuating) extrapyramidal disorders, in particular Parkinsonism dystonia or more general "athetoid cerebral palsy", and vegetative disturbances. A severe epileptic encephalopathy and progressive mental retardation may be present. Neuroimaging findings do not show specific lesions. Determinations of folates and organic acids in CSF appear at present only warrantable individually in special constellations, e.g. classical clinical findings and disease course suggestive of glutaryl-CoA dehydrogenase deficiency with repeated negative quantitative analyses of organic acids in urine. The diagnosis of disorders, which require specific analyses of CSF, can only be achieved by conscious diagnostic decisions based on a concept of the respective disease and repeated scrupolous expert clinical evaluation aided by an array of investigations in blood and urine as well as neuroimaging findings. No single one investigation in CSF can serve as a "selective screening" test. A growing awareness of these disorders is needed and should lead to increased and earlier diagnosis of patients through fewer rather than more lumbar punctures.