Cannula tubing (1.6 mm external, 1 mm internal diameter) was fabricated from medical grade polyurethane containing 2% 2,4,4'-trichloro-2'-hydroxydiphenyl ether ('Irgasan', Ciba-Geigy). When shaken with 10 ml of phosphate-buffered saline an equilibrium was set up with only traces of 'Irgasan' in the aqueous phase. When phosphate-buffered saline flowed through the tubing, 0.26 mg of 'Irgasan' per g of tubing per day was eluted, and this was not detectably increased by buffers of pH 3 or pH 8.5. However, a 20% lipid emulsion eluted 6 mg per g of tubing per day with a flow rate of 250 ml per day and 11 mg with a flow rate of 11 per day. These results scaled up show that a 7.11 g Hickman catheter would lose 78 mg of Irgasan in the first day with Intralipid, compared to 1.8 mg with PBS. If elution by tissue fluids (including blood) does not exceed that by Intralipid then comparison with available toxicological data suggests an adequate margin of safety for adults, but not for premature babies.