We have investigated the effect of the G/A substitution in the promoter region of the apolipoprotein (apo) AI gene (-75 bp) on plasma lipid, lipoprotein and apolipoprotein levels in a sample of 204 children from central Italy. The subjects included 111 boys and 93 girls, aged 8-11 years old. The frequency of the A allele was 0.19 in the total sample, and 0.21 and 0.17 in boys and girls, respectively. Using analysis of variance, we found the G/A substitution was significantly associated with plasma levels of total cholesterol, LDL cholesterol, apo B, and apo AI in boys, accounting for 7.0, 4.2, 5.3, and 4.3% of the sample variance, respectively. Individuals with an A allele had higher mean levels of these lipid traits than individuals homozygous for the G allele. A dietary intervention study had been carried out in a subset of these children, and the effect of the G/A substitution on plasma apo AI levels remained when boys changed to a low fat low cholesterol diet. However, no significant association was observed in girls between any of the lipid traits and the G/A genotypes. We have previously reported in this sample of children that the two polymorphisms detected with restriction enzyme PvuII, with variable sites in the first intron of the apo CIII gene (Pvu II-CIII) and the apo CIII-AIV intergenic region (Pvu II-AIV), were associated with significant differences on plasma apo AI levels. We found that the association reached statistical significance in boys only in this study. Taking these three polymorphisms together, the effects on plasma apo AI levels were additive in boys, accounting for 20.0% of the sample variance. Boys having the genotype GG/V-V+ of the G/A substitution and the PvuII-AIV RFLP had mean apo AI levels 36 mg/dl lower than boys with the genotype GA + AA/V-V-. In girls, however, there was evidence of significant interaction of effects between the PvuII-AIV RFLP and the G/A substitution (P < 0.04), with the A allele being associated with higher levels of plasma apo AI only in girls having the rare allele (V+) of the PvuII-AIV RFLP. We conclude that genetic variation at the apo AI-CIII-AIV gene cluster is having a major impact on the determination of plasma apo AI levels in this sample of young boys, with additive effects due to functional changes at several places in this gene cluster detected directly (G/A) or in allelic association with the PvuII-CIII and PvuII-AIV polymorphisms.(ABSTRACT TRUNCATED AT 400 WORDS)