Although a subarachnoid-pleural fistula (SPF) is a rare entity, at least 50 cases have been described in the English literature since 1959. Most cases are due to thoracic injuries or surgical complications. The subsequent meningeal tear results in the establishment of a communication between the spinal subarachnoid space and the pleural cavity. The negative pleural pressure during inspiration draws cerebrospinal fluid (CSF) into the pleural cavity, often resulting in a massive pleural effusion because approximately 20 mL CSF is produced each hour. CSF production is increased in patients with choroid plexus papillomas as well as with caffeine intake. Decreased CSF production has been associated with aging, the administration of acetazolamide, or acute ventriculitis. Patients may prefer the supine position because of orthostatic headache, but it is unclear whether bed rest results in a reduction of CSF production. The pleural effusion should be transudative with very low total protein concentration, much less than 1 g/dL when SPF is not complicated by trauma, infection, or bleeding. In our case, the relatively high total protein in the pleural fluid was presumed to be secondary to surgical procedures including thoracotomy and chest tube placement. The diagnosis has been made previously by exploratory surgery, myelography, or nuclide scintimyelography when SPF was suspected. However, the diagnosis can be difficult and delayed because of the rarity of the entity, the invasive nature of the diagnostic tests, and the complex intrathoracic pathology, such as hemothorax, pneumothorax, chylothorax, or empyema, and it may delay in presentation up to 6 years after trauma. A delayed diagnosis may result in life-threatening complications such as meningitis or pneumocephalus. The natural course of SPF probably depends on the size of the dural tear, and most patients require a surgical repair. However, spontaneous sealing of the CSF leak has been reported in approximately 20% of cases. In our case, a moderate left pleural fluid has remained without symptoms for 3 months after the last thoracentesis, and no further imaging study, such as myelography or nuclide scintimyelography, or surgical repair has been indicated.