To study the effect of in utero ventricular decompression of hydrocephalus on brain development and prognosis, and to evaluate the function and possible complications of different shunt designs, we created fetal hydrocephalus in 28 fetal lambs and 17 fetal monkeys by injecting kaolin into the cisterna magna during the third trimester. One fetal lamb had indwelling intracranial subdural and amniotic cavity pressure catheters placed to study serial changes in the relationship of intracranial pressure (ICP) and amniotic fluid pressure (AFP) from the time of injection until term. Twenty hydrocephalic fetal lambs underwent ventricular decompression--ventriculoamniotic (V-A), N = 10; ventriculo-right atrial (V-RA), N = 9; and ventriculo-pleural (V-PL), N = 1--21 to 25 days after the kaolin was injected; seven hydrocephalic fetuses were left unshunted as controls (CON). Eight fetal monkeys underwent V-A decompression 14 to 21 days after kaolin injection; nine were left unshunted as controls. All animals were delivered by cesarean section near term, assessed for viability and their brains examined grossly and microscopically. After kaolin injection in the fetal lamb, ICP (r = 0.94) and ICP minus AFP (r = 0.93) rose in a linear fashion, while AFP showed no trend (r = 0.22). All unshunted newborn lambs and monkeys had split sutures, dilated ventricles, and thinned cortical mantle. Most shunted lambs showed anatomic improvement with decreased head circumference, overriding sutures, normal-sized ventricles, and improved survival. However, histopathology revealed marked white matter destruction. In contrast, most shunted monkeys showed little anatomic improvement. Both shunted and unshunted monkey brains showed a severe inflammatory ventriculitis. In both lambs and monkeys, shunting was associated with a variety of complications including subdural hematoma, subdural hygroma, shunt infection, shunt occlusion, and improper shunt tip placement. In this experimental model, in utero decompression of obstructive hydrocephalus improves overall survival, improves gross ventriculomegaly, does not improve histopathologic brain damage, and is associated with significant complications. The choice of the optimal shunt design and the effect on postnatal neurologic function requires further study. This work emphasizes the need for continuing research in an animal model prior to human application.