Several mouse mutations disturb the embryonic process of neurulation, yielding neural tube defects. Analysis of the mutations offers the most feasible approach to understanding the aetiology and pathogenesis of human neural tube defects. Interactions between the non-allelic mutant genes and between several of the mutant genes and modifying genes in the genetic background modulate the frequency and severity of the defects that develop. Environmental factors interact with the genetic predisposition either to increase or to decrease the incidence of defects. The gene loci corresponding to two of the mutations, splotch (Sp) and extra toes (Xt), have been identified as those encoding the transcription factors Pax-3 and Gli3, respectively; their human homologues are associated with Waardenburg type I syndrome and Greig's cephalopolysyndactyly. Embryological analysis reveals that several of the mutations disturb the process of neural tube closure at the posterior neuropore (in the lumbosacral region), yielding spina bifida and/or tail defects. The different mutations appear to achieve this developmental end-point by different underlying mechanisms. In curly tail (ct), non-neural tissues proliferate abnormally slowly causing ventral curvature of the neuropore region and inhibiting neural tube closure. Neural tube defects can be prevented in cultured ct/ct embryos by experimentally correcting either the proliferative imbalance or the ventral curvature. In Sp the primary defect appears to reside in the neuroepithelium. A combination of genetic analysis, gene cloning and experimental embryology is revealing that neural tube defects in mice and, by implication, in humans are a developmentally heterogeneous group of malformations.