In a multicentre prospective study, 124 cases of meningococcal disease were classified into the clinical categories, meningitis alone (n =15), meningitis and septicaemia (n= 79) and septicaemia alone (n =30). A further 60 children referred with other illnesses served as controls. Serial measurements of serum C‐reactive protein (admission, day 1, day 2, days 5–7) were compared. Children with septicaemia had significantly lower C‐reactive protein levels on admission than those with meningitis alone or meningitis and septicaemia which were unexplained by differences in the duration of the presenting illness or severity of the disease. Within each clinical category of meningococcal disease, significant changes in C‐reactive protein concentration occurred during the course of the disease. Four control children had other types of septic meningitis: admission C‐reactive protein concentrations did not differ from those with meningitis or meningitis and septicaemia, but were significantly higher than those with septicaemia alone. The other 56 patients had a significantly lower admission C‐reactive protein concentration compared with all cases of meningococcal disease. For the diagnosis of meningococcal disease, admission C‐reactive protein levels of 40 mg/l had a sensitivity of 79%, specificity of 80% and positive predictive value of 87%. For the prognostic prediction of death in meningococcal disease (or meningococcal disease with shock) CRP<100 mg/l on admission had a sensitivity of 69% (69%), specificity of 50% (56%) and positive predictive value of 18% (53%). In children with suspected meningococcal disease, serum C‐reactive protein, measured on admission, has diagnostic value but not prognostic value. The C‐reactive protein concentration on admission varies with the clinical category of meningococcal disease and t