A randomized crossover study with eight dogs was carried out to determine whether the in vitro difference observed in the delivery of nitroglycerin using polyvinyl chloride (PVC) and polyethylene (PE) infusion sets would result in a measurable difference in steady-state plasma concentration. At the same apparent infusion rate of 40 micrograms/min, PVC infusion sets produced steady-state plasma nitroglycerin concentrations that were only 40% of those generated with PE sets (p less than 0.0001). Attainment of steady-state drug levels appeared more rapid with the PE administration sets, but in three out of eight animals, these infusion systems appeared to produce a "bolus" dosing effect. PE sets did not reduce the variability in plasma nitroglycerin concentration observed with PVC sets.