Though a genetic basis for non-insulin-dependent diabetes mellitus (NI
DDM) is clear, the likely mode of inheritance is not known. The segreg
ation of NIDDM was studied in 64 nuclear South Indian pedigrees (449 i
ndividuals) ascertained through an affected proband having both parent
s and more than 1 sibling alive and available for oral glucose toleran
ce testing. A high proportion of parents were found to be of abnormal
glucose tolerance [89 of 128 (70 %) diabetic and 11 of 128 (9 %) impai
red]. Complex segregation analysis was performed using (1) POINTER whi
ch implements the mixed model and distinguishes major gene, multifacto
rial and nontransmitted environmental contributions to affection and (
2) COMDS which implements an oligogenic model with major gene, modifie
r gene and environmental contributions to a) affection and b) diathesi
s (an ordered polychotomy amongst non-affected family members, based o
n 2-h plasma glucose level). Using POINTER, there was no formal suppor
t for a major gene and the most parsimonious solutions were achieved w
ith multifactorial models Using COMDS, we found i) significant improve
ments in models when information on glucose levels in nondiabetic fami
ly members (diathesis) was included, ii) support for segregation of a
diallelic gene as well as background familial resemblance, and iii) un
der the best-supported model, this diallelic locus featured incomplete
dominance (d = 0.8) and a disease-predisposing allele frequency of 14
%. In South Indians, segregation of NIDDM is inadequately described b
y simple major gene models: more complex models provide more satisfact
ory descriptions. This finding, if applicable in other populations, ha
s important implications for the search for diabetes-susceptibility ge
nes.