The purpose of this study was to analyse the use and relative complication rates of Hickman lines inserted by angiographic and surgical routes in patients with solid tumours. Sixty-one patients (median age 55 years; range 21-81) with solid tumours had 71 Hickman catheters inserted. Fifty-three patients had 58 catheters inserted angiographically and eight patients had 13 catheters inserted surgically. Indications were as follows: for administration of infusional chemotherapy (48 patients), poor venous access (nine patients), needle phobia (four patients). The catheters remained in situ for 7981 days (median 105; range 10-750). Immediate complications occurred only in angiographically placed catheters (ten pneumothoraces, one tachydysrhythmia). Pneumothorax was associated with poor nutritional status but not with age, side of insertion or catheter gauge. The median body mass indices (BMIs) were 19.65 kg/m2 and 22.8 kg/m2 in those with and without pneumothorax respectively (P = 0.0045). Five of six patients with BMI < 19 kg/m2 developed a pneumothorax, compared with five of 52 with BMI > 19 kg/m2 (P = 0.0003). The overall sepsis rate was 2.38 per 1000 catheter days. Systemic sepsis was significantly increased for surgically inserted lines (3.94 versus 0.60 per 1000 catheter days; P = 0.001), although there was no significant difference in superficial sepsis (2.36 versus 1.04 per 1000 catheter days; P = 0.22). We conclude that Hickman catheters provide safe, long-term venous access in patients with solid tumours. Angiographic insertion carries a greater risk of acute complications in patients of poor nutritional status but a significantly lower risk of late systemic infective complications.(ABSTRACT TRUNCATED AT 250 WORDS)