Serial procalcitonin is reported to be useful to titrate duration of antibiotic therapy in the non critically ill patient with pneumonia. The aim of this study was to examine the relationship between antibiotic therapy and serial serum procalcitonin concentrations in a cohort of critically ill septic patients and examine for any differences between culture positive (CP) and culture negative (CN) sepsis. Seventy-five critically ill patients with suspected sepsis were enrolled in this prospective observational study. Serial procalcitonin and C-reactive protein assays were measured on days one, three, five, seven, 10 and 14. The mean duration of antibiotic therapy was similar in the two groups (10.4 +/- 5.1 (CP) vs. 8.4 +/- 5.1 (CN) days, P = 0.09). Serum procalcitonin concentrations were significantly higher at baseline in the CP than the CN group (14.9 +/- 22.9 vs. 6.8 +/- 21.5 ng/ml, P = 0.04). During the study period, serum concentrations of procalcitonin and C-reactive protein declined in both groups. Serum procalcitonin consistently remained higher in the CP group (P < 0.05) and did not return to normal values. In the CN group, procalcitonin concentrations fell below 0.5 only on day 10. There was no significant difference in C-reactive protein profile between the two groups. Four patients in the CP group (11%) had relapse of sepsis. The mean procalcitonins in the relapsed subgroup were lower than those in the remission subgroup (P = 0.02). Therapy for proven or presumed infections was associated with declining serum procalcitonin and C-reactive protein in critically ill septic patients. The marked variability and overlap in plasma profile of these markers between CP and CN sepsis makes it difficult to define a nadir plasma concentration at which one can recommend discontinuation of antibiotic therapy.