To conduct a case-control study to investigate whether there are independent tumour necrosis factor alpha (TNFalpha) or lymphotoxin alpha (LTalpha) haplotype associations with SLE or with any of the major serological subsets of SLE.157 patients with SLE were genotyped for HLA-DRB1, HLA-DQB1, TNFalpha, and LTalpha alleles by polymerase chain reaction and compared with 245 normal white controls. For TNFalpha, six single nucleotide polymorphisms (SNPs) at positions -1031, -863, -857, -308, -238, and +488 and for LTalpha three SNPs at positions +720, +365, and +249 were studied to assign six TNFalpha haplotypes (TNF1-6) and four LTalpha haplotypes (LTA1-4). All SLE patients had full serological profiles on serial samples.The most significant association with SLE overall was with HLA-DR3 (p<0.001; odds ratio (OR) = 2.5 (95% confidence interval, 1.6 to 3.8)) and the extended haplotype HLA-DQB1*0201;DRB1*0301;TNF2;LTA2 (p<0.001; OR = 2.3 (1.4 to 3.7)). Associations were strongest in the anti-La positive group (13%) of SLE patients (HLA-DR3, OR = 71 (9 to 539); HLA-DQB1*0201, OR = 35 (5 to 267); TNF2, OR = 10 (2.8 to 36), and LTA2, OR = 4.9 (1.1 to 21)). There was an increase in the HLA-DR2 associated extended haplotype (HLA-DQB1*0602;DRB1*1501;TNF1;LTA1) in patients with anti-Ro in the absence of anti-La (p<0.005; OR = 3.9 (1.5 to 10)). The HLA-DR7 extended haplotype (HLA-DQB1*0303;DRB1*0701/2;TNF5;LTA3) was decreased in SLE overall (p<0.02; OR = 0.2 (0.05 to 0.8)).The strongest association in this predominantly white population with SLE was between HLA-DR3 and anti-La, which seemed to account for any associations with TNFalpha alleles on an extended DR3 haplotype.