OBJECTIVE: Brain biopsy is standard clinical practice when CNS malignancy is suspected. Its value has not otherwise been clearly established. We reviewed the indications for, complications and outcome of diagnostic brain biopsies performed between 2003 and 2008 in a single UK centre. METHODS: Subjects were retrospectively identified using theatre log books and histopathology reports. Case records were analysed by a neurologist and neurosurgeon. Cases were excluded when the pre-operative diagnosis was clearly malignancy or infection. RESULTS: Of all (432) brain biopsies performed, 56 were performed in 52 patients with cryptogenic neurological disease. There were no permanent deficits or deaths. Histopathological reports were classified as definitive (45%), suggestive (20%) or non-diagnostic (36%). Brain biopsy made an immediate contribution to determination of diagnosis in 55% (31 of 56) and a confident diagnosis was eventually made in 40 of 52 patients (77%). Management was altered as a consequence of biopsy in 63%. Successful biopsy of a radiologically identified target increased the proportion of biopsies considered diagnostic to 78% (odds ratio 8.9) whereas non-targeted biopsy was non-diagnostic in 71%. Although a significant proportion of patients died or had progressive disease, this was not uniformly the case; 31% stabilised and 27% improved. CONCLUSION: We present the highest reported frequency of brain biopsy for cryptogenic neurological disease. The risk associated with the procedure was low and the biopsy results impacted significantly upon diagnosis and management. We therefore propose that the procedure should no longer be considered one of last resort.