The possibility that specific environmental factors such as smoking and estrogen use modify the genetic influences (gene–environment interaction) on bone mineral density (BMD) has not been explored in genetic epidemiological studies such as twin studies. The aim of this study was to look for evidence of gene–environment interaction in BMD determination by analyzing data collected on a large number of healthy female twins. BMD of the hip, distal forearm and lumbar spine were measured by dual-energy X-ray absorptiometry on 287 identical and 265 nonidentical volunteer female twin pairs. The environmental factors examined were hormone replacement therapy (HRT) and smoking. In genetic modeling analysis using path analysis, there was evidence of ‘HRT-specific’ genetic component of BMD variance at the forearm (50% of total variance) but not at the hip. At the lumbar spine the magnitude of the genetic component of variance in HRT users (> 60-month HRT use) was less than the genetic component of variance for little or no exposure to HRT (48% vs. 84%). There was no evidence of gene–environment interaction for smoking. The main evidence for gene–environment interaction was the finding that forearm BMD variance was influenced by a significant HRT-specific genetic component. There was also evidence that in HRT users, the genetic component of total variance for lumbar BMD was lower.