Decompressive craniectomy has been advocated as a treatment for the cerebral edema associated with massive head injury. Although craniectomy has been successful in lowering intracranial pressure after head injury, a comparison of computerized tomographic scans of comparable patients with traumatic brain edema treated by medical means or decompressive craniectomy showed that bony decompression resulted in apparent exacerbation of edema. To investigate the possibility of enhancement of brain edema by craniectomy, we produced standardized cold lesions in the brains of 10 dogs. Five animals served as controls. In the other 5 animals we performed large decompressive craniectomies after lesioning. Physiological parameters were comparable in both groups. The dogs were killed 8 hours after lesioning. After fixation, their brains were cut into 1-mm-thick slices. We used an image analysis facility built around a PDP 11/105 computer to measure the volume of edema as outlined by Evans blue staining. The mean volume of the brain edema in the control animals was 0.27 +/- 0.19 ml. Mean edema volume was over 7 times greater in craniectomized animals (1.96 +/- 1.89 ml). This difference is statistically significant (p less than 0.05). The driving force for the formation of edema fluid is the difference between intravascular and interstitial presssure. Decompression of the brain by bone removal probably results in a reduction of interstitial fluid pressure and edema enhancement. The clinical literature contains no evidence that craniectomy decreases the morbidity or mortality of human head injury. In view of our experimental findings, this is not surprising. Indeed, pathological evidence indicates that severe edema (such as that accentuated by craniectomy) may produce permanent changes in the neuropil.