Abstract Objective Recent studies indicate that removal of the primary tumour may have a beneficial effect on mortality risk of patients with primary distant metastatic breast cancer (stage IV), although most of them did not rule out confounding by the presence of co-morbidity. In this retrospective study the impact of surgical resection of the primary tumour on the survival of patients with primary distant metastatic disease is investigated, taking into account the presence of co-morbidity and other potential confounders. Methods Between 1993 and 2004, 15 769 patients with breast cancer were diagnosed in the south of the Netherlands. This study included the 728 patients with distant metastatic disease at initial presentation, which was 5% of all patients. Of them, 40% had surgery of the primary tumour. Follow-up was carried out until 1 July 2006. Results Median survival of the patients who had surgery of their primary tumour was significantly longer than for the patients who did not have surgery (31 vs. 14 months). The 5-year survival rates were 24.5% and 13.1%, respectively ( p < 0.0001). In a multivariable Cox regression analysis, adjusting for age, period of diagnosis, T-classification, number of metastatic sites, co-morbidity, use of loco-regional radiotherapy and use of systemic therapy, surgery appeared to be an independent prognostic factor for overall survival (HR = 0.62; 95% CI 0.51–0.76). Conclusion Removal of the primary tumour in patients with primary distant metastatic disease was associated with a reduction of the mortality risk of around 40%. The association was independent of age, presence of co-morbidity and other potential confounders, but a randomized controlled trial will be needed to rule out residual confounding.