Scientists have long puzzled over nerve regeneration. Damaged nerves in the arms or legs often repair themselves after injury, but those in the brain or spine rarely do. Research in the past 2 decades has suggested that GAP-43, a compound in the growth-associated protein family, influences nerve regrowth. But in laboratory trials, injured animals induced to produce extra amounts of this compound show only meager signs of spinal cord repair. Another compound under investigation, a cortical-actin-associated protein, CAP-23, has also performed feebly.