The paper describes associated malformations in infants born with neural tube defects (N ϭ 3,809) from three large malformation registers and in fetuses aborted because of a diagnosed neural tube defect (N ϭ 748) from two of the registers. In infants, upper spina bifida and encephalocele are more often associated with non-neural malformations than anencephaly or lower spina bifida. Aborted fetuses with spina bifida or encephalocele have associated malformations registered more often than infants with those neural tube defects, but the opposite is true for anencephaly. The degree of detail of the investigation of an aborted specimen or a perinatally dead infant will contribute to such differences but they can also depend on the fact that prenatal detection may be facilitated by the simultaneous presence of other malformations like body wall defects. Also, fetuses with many malformations may be more prone to abort spontaneously late in pregnancy. Variable prenatal diagnosis may, therefore, explain population differences in the pattern of associated malformations. The type of associated malformation differs with the level of the neural tube defect: this could be due to different causal mechanisms or be a question of cranio-caudal level and/or timing. For limb reduction defects, however, we did not find any association between upper limb and upper neural tube defects or lower limb and lower neural tube defects. These findings together with other epidemiological data support the idea that upper and lower neural tube defects may have different significance in epidemiological studies and should be treated separately.