We examine the interaction between stressful life events (SLE) and genotypes for the length polymorphism of the serotonin receptor gene (5HTTLPR) on risk of depression. We hypothesize that if the interaction is real, monozygotic twin pairs (MZT) homozygous for the short allele (SS) will have a greater within pair variance in depression measures than MZT homozygous for the long allele (LL), as a reflection of their increased sensitivity to unknown environmental risk factors. Telephone interviews were used to assess symptoms of depression and suicidality on 824 MZT. Rather than using the interview items to calculate sum scores or allocate diagnostic classes we use Item Response Theory to model the contribution of each item to each individual's underlying liability to depression. SLE were also measured on the MZT assessed by mailed questionnaire on average 3.8 years previously, and these were used in follow-up analyses. We find no evidence for significant differences in within pair variance between 5HTTLPR genotypic classes and so can provide no support for interaction between these genotypes and the environment. The use of MZT provides a novel framework for examining genotype × environment interaction in the absence of measures on SLE.