In a controlled pilot study of 32 critically ill patients, we have attempted prospectively to identify laboratory variables which can be used to select and monitor patients on antithrombin (AT) therapy. Patients with plasma AT levels less than 70% of normal were randomized to receive (AT group) or not to receive AT concentrate (non-AT group). The groups did not differ in median age, sex, median APACHE II and TISS scores, number of days spent in the Intensive Care Unit or mortality rate. At the time of inclusion all patients had activated coagulation and fibrinolysis demonstrated as high levels of soluble fibrin, thrombin-antithrombin complexes and fibrin-D-dimers (twice, four and ten times the upper reference range, respectively). In the AT group these levels decreased faster and the prothrombin complex concentration increased more rapidly to normal (i.e. the prothrombin time decreased). The level of C-reactive protein which was high in both groups on inclusion (139 and 98 mg/l, respectively) decreased by 40% in the AT group but did not change in the control group. Our study indicates that laboratory variables normalize faster in seriously ill patients who have activated coagulation and fibrinolysis when they receive AT concentrate and that the variables mentioned above seem to be useful for monitoring the treatment.