BACKGROUND Despite affecting approximately one-quarter of all patients undergoing axillary lymph node dissection, the pathophysiology of breast cancer-related lymphoedema (BCRL) remains poorly understood. More extensive locoregional treatment and higher body mass index have long been identified as major risk factors. This study aimed to identify risk factors for BCRL with a specific focus on the potential impact of chemotherapy on the risk of BCRL. METHODS This was a retrospective analysis of a cohort of consecutive patients with breast cancer treated at a major London regional teaching hospital between 1 January 2010 and 31 December 2012. All patients had node-positive disease and underwent axillary lymph node dissection. Data regarding tumour-, patient- and treatment-related characteristics were collected prospectively. The diagnosis of BCRL was based on both subjective and objective criteria. Multivariable Cox proportional hazards regression was used to assess the association between treatment and risk of BCRL. RESULTS Some 27.1 per cent of all patients (74 of 273) developed BCRL over the study period. Administration of taxanes showed a strong association with the development of BCRL, as 52 (33.5 per cent) of 155 patients who received taxanes developed BCRL. Multivariable Cox regression analysis demonstrated that patients who received taxanes were nearly three times more likely to develop BCRL than patients who had no chemotherapy (hazard ratio 2.82, 95 per cent c.i. 1.31 to 6.06). No such increase was observed when taxanes were administered in the neoadjuvant setting. CONCLUSIONS The present findings suggest that adjuvant taxanes play a key role in the development of BCRL after surgery. This may support the use of taxanes in a neoadjuvant rather than adjuvant setting.