Latex agglutination (LA) of capsular polysaccharide bacterial antigen is a frequently performed laboratory procedure, but its use is controversial. To assess the clinical utility of this test, we reviewed all LA tests performed over a 10-month period at two sites, a major university-based referral center and a private specialty pediatric hospital. Samples were assayed either individually or as a panel for the group B streptococcus, Streptococcus pneumoniae, Haemophilus influenzae, and three sets of Neisseria meningitidis serogroups (A and Y, C and W135, and B and Escherichia coli K1). Of 5,169 assays performed on 1,268 clinical samples (786 urine and 478 cerebrospinal fluid, 3 pleural fluid, and 1 synovial fluid sample), 57 (1.1\%) were positive, including 1.7\% of urine and 0.3\% of cerebrospinal fluid samples. All LA true-positive cerebrospinal fluid samples showed the causative microorganisms by Gram stain. Detailed chart review of these 57 positive samples showed that the LA result was false-positive in 31 (54\%), true-positive in 22 (38\%), and indeterminate in 4 (7\%) samples. Therapy was not altered on the basis of any of the true-positive LA results. The 31 false-positive results led to additional cost, prolonged hospitalization, and some clinical complications. Total patient charges were \$175,000 (\$7,954 per true-positive), with no detectable clinical benefit. Our retrospective study does not support the current use of LA for rapid antigen detection. What, if any, specific indications exist for this test remain to be elucidated.