This study made a longterm (72 hours) evaluation of the efficacy and possible side-effects of transdermal delivery of fentanyl (TTS-system) for post-operative pain relief. The study was double-blind, placebo-controlled with either a TTS-system delivering fentanyl 100 micrograms.h-1 and rescue analgesic on demand or a placebo system and analgesic on demand. Analgesic consumption, pain, general satisfaction, respiratory rate, and levels of SpO2 and tcCO2 (pulse oximetry and transcutaneous CO2 measuring) were evaluated. Recruitment was stopped after enrolment of 24 patients, on safety grounds. The Fentanyl group was more satisfied with postoperative pain relief (P = 0.008); they had a lower analgesic demand (P < 0.05) but also a lower respiratory rate (P < 0.05) and a higher level of tcCO2 23 hours after application (P < 0.05). There were three cases (25%) of increased PaCO2 (> 6.5 kPa) in the Placebo group but without low PaO2 levels, sedation or bradypnoea. Conversely, there were three cases (33%) in the Fentanyl group with bradypnoea (< 10 breaths/minute), two without influence on PaO2 or PaCO2, but one (no. 24) with bradypnoea, heavy sedation, a marked decrease in PaO2 (5.8 kPa) and increased PaCO2 (7.5 kPa). These findings terminated the study. The 100 micrograms transdermal fentanyl system is agreeable to the patients, but apparently too potent for routine postoperative pain relief due to a risk of respiratory depression. Respiratory frequency can not be relied upon as sole indicator of insufficient respiration.