The clinical course of 72 septicemic episodes or focal severe bacterial infections was monitored by daily measurements of serum C-reactive protein (CRP) in 59 children beyond the neonatal period, 19 of whom were immunocompromised. CRP was determined quantitatively by an immunoturbidimetric method from a finger prick sample until either clinical recovery occurred and antimicrobial therapy was discontinued or until the death of the patient. The primarily elevated CRP levels (greater than or equal to 20 mg/l) usually increased about for a day but then decreased rapidly, provided the patient recovered uneventfully. If not, CRP remained at a high level or reincreased after transient decrease. Behaviour of CRP was not affected by the immunologic status of the patient. This property makes CRP especially useful in immunocompromised patients in whom other commonly used laboratory parameters may fail.