In November‐December 1977 an epidemic of bacteræmia due toP. cepaciawas observed in Odense, Denmark (nine patients), and in Nijmegen, Holland (seven patients). All patients recovered. The epidemic was traced to intrinsic contamination of two batches of the anæsthetic fentanyl. All isolates from the patients and from the two batches belonged to the same biotype, had identical sensitivity patterns, and identical antigens. TheP. cepaciastrain differed from stock strains in being able to grow in two passages in methyl‐p‐hydroxi‐benzoate, 0.5 mg/ml, which promoted the growth of the microorganism; inocula of 2–20 cfu were sufficient to initiate growth in the drug or preservative. These facts indicate the inadvisability of using p‐hydroxi‐benzoates as preservatives in vials. The strain was inhibited at temperatures above 38.5°C, corresponding to the recovery of the patients after a period with fever above 39°C. Fourteen out of 15 patients examined had agglutinin titres ≥ 320, while 36 blood donors had titres<40. Of 12 patients with postoperative fever in the same period whose blood cultures did not yieldP. cepacia, t