The purpose of this study was to identify risk factors for pneumothorax during percutaneous subclavian Hickman line insertion in patients with haematological (HT) or solid tumours (ST). One hundred and twelve patients (55 HT, 57 ST) had 132 subclavian Hickman lines inserted under fluoroscopic control. Lines were inserted on the left on 116 occasions and the right in 16. Thirty-five single, 29 double and 68 triple lumen catheters were inserted. Variables included in the analysis were age, sex, side of insertion, catheter gauge and nutritional status as indicated by the body mass index (BMI kg/m2). Both univariate and logistic regression analyses were performed. There were ten pneumothoraces, all occurring in patients with STs. Univariate analysis revealed that patients with STs were older (median age 59 versus 36 years; P = 0.0001) and more cachectic (median BMI 21.9 versus 24.2 kg/m2; P = 0.03) than those with HTs, and the patients experiencing pneumothorax were older (median age 57 versus 44 years; P < 0.01) and more cachectic (median BMI 19.6 versus 24.0 kg/m2; P = 0.004) than those undergoing uncomplicated procedures. For patients with a BMI < 19 kg/m2 versus those with a BMI > 19 kg/m2, the pneumothorax rate was 5/8 versus 5/124 (P < 0.00001). Logistic regression analysis confirmed the significantly older age of the patients with STs (P < 0.001) and of those experiencing pneumothorax (P = 0.02). This analysis also confirmed the lower BMI of patients experiencing pneumothorax (P = 0.002). The patients' sex, the side of line insertion and the catheter gauge were not associated with pneumothorax on either the univariate or logistic regression analyses. From these data we conclude that pneumothorax complicating a subclavian Hickman line is significantly more likely in elderly patients with a low BMI. Such patients are more likely to have STs than HTs. In patients with a BMI < 19 kg/m2, the subclavian route should be used with caution.