Two simple and low-priced kits to detect bacterial antigens in cerebrospinal fluid (CSF) specimens from patients with suspected meningitis have recently become available. Methods employed by these kits are staphylococcal coagglutination (COA) and latex agglutination (LA). The COA "Phadebact CSF-kit" (Pharmacia) detects Haemophilus influenzae type b (HIb), Neisseria meningitidis (NM) groups A, B, C, Y and W 135 and Streptococcus pneumoniae (SP), whereas the LA "Slidex Méningite-kit" (BioMérieux) includes HIb, NM A, NM C and SP. These two diagnostic tests were compared with the standard methods for analyzing CSF specimens from children with suspected meningitis. A total of 336 CSF specimens were tested. Forty-three were from children with bacterial meningitis due to HIb, NM or SP obtained before antimicrobial therapy. Thirty-four of them (79%) were correctly detected by COA, 23 (54%) by LA, and 42 (97%) by microscopic analysis of stained smears. There were 65 CSF samples of such meningitis cases obtained after initiation of antimicrobial therapy: COA detected 26 (40%), LA 17 (26%) and staining 26 (40%). With culture-positive CSF specimens from patients before therapy, the best sensitivities were obtained with LA for SP (100%) and COA for HIb (87%), whereas the results for NM antigen detection were only 43% with COA and 0% with LA (no reagent against NM B). Cross-reactions with other bacterial antigens were frequent with COA (26%) and rare with LA (2%), resulting in false-positive findings in 6 of the 66 positive tests with COA (9%) but in none of the 40 positive tests with LA (0%). One of the 48 specimens from aseptic meningitis cases was false-positive for HIb by COA, whereas all 172 normal specimens were negative with both methods. With 5 culture-negative CSF specimens from patients with unequivocal purulent meningitis COA detected bacterial antigen in 5 and LA in 2. CONCLUSIONS The two kits evaluated cannot replace standard methods for analyzing CSF specimens, in view of insufficient sensitivity (mainly LA) or frequent cross-reactions and false-positive values (COA). However, due to their simplicity, rapidity, and possible identification of bacterial pathogens after initiation of antimicrobial therapy, these kits are recommended as an useful addition to standard methods.