A high degree of suspicion is required to establish the diagnosis of subarachnoid-pleural fistula. Often, symptoms related to the effusion such as chest pain, dyspnea, or fever may be minimal or overshadowed by concomitant injuries, and often the diagnosis is delayed. In addition to the almost universal finding of severe spinal cord injury, rare clues to the diagnosis may include pneumocephalus or meningitis. Symptoms that suggest a CSF leak include headache, nausea, and vomiting. The findings of a chest radiograph initially may be normal. Effusions range from small to massive, depending on the size and duration of the fistula. Mediastinal widening also may be seen, and, once aortic injury is excluded, a diagnosis of subarachnoid-mediastinal or subarachnoid-pleural fistula should be considered. Pleural fluid in a subarachnoid-pleural fistula has been described most often as being clear with a low nucleated cell count, having a glucose equivalent to serum, and having a low protein content in the transudative range.6–7 However, there may be concomitant pleural processes making pleural fluid analysis problematic, as in the present patient who initially had a hemopneumothorax. Evidence that the hemothorax significantly influenced the pleural fluid analysis may be inferred by the eosinophilia noted in the pleural fluid sample 16 days after the injury (Table 1). Blood in the pleural space is a known cause of an exudative, eosinophilic pleural effusion. The peak pleural fluid eosinophil count tends to occur approximately 7 to 14 days following the entry of blood into the pleural space.8 During his hospital course, the patient developed fever, chest pain, and worsening oxygenation with a radiographic opacity that was suggestive of pneumonia with parapneumonic effusion. It is postulated that, for these reasons, the pleural fluid analysis was most consistent with an exudate rather than the clear transudate described in the literature. Shannon and colleagues9reported a similar case of a 27-year-old stabbing victim who presented with a hemothorax and persistent pleural drainage that was diagnosed as a subarachnoid-pleural fistula. The pleural fluid continued to have a high protein concentration 2 months after the closure of the pleural fistula. As in our case, electrophoresis for β2-transferrin failed to identify CSF in the pleural fluid. β2-transferrin has been widely used to identify CSF leakage following head trauma, as it is found only in CSF and inner ear perilymph fluid. The test is both sensitive (100%) and specific (95%) for CSF leaks due to head trauma.10 β2-transferrin never has been studied in the setting of a subarachnoid-pleural fistula, and we postulate that concomitant pleural processes may lead to a false-negative result.