The Cand-Tec Candida detection system and enzyme immunoassay for serum mannan were retrospectively compared in a controlled collaborative evaluation of antigen detection in 32 patients with candidiasis proven by biopsy or culture from a normally sterile site and with sera drawn within 7 days of inclusion. With a threshold titer of 1/8, which excluded false-positive results in 17 hospitalized patients without candidiasis, sensitivities for all 32 patients with candidiasis were 44% for the Cand-Tec assay and 17% for the enzyme immunoassay. Both assays provided greater sensitivity when sera were drawn within 24 h of inclusion in the study and in the category of patients with invasive candidiasis (57% by Cand-Tec and 33% by enzyme immunoassay). The Cand-Tec assay gave false-positive results (titer, greater than or equal to 1/8) in 4 of 6 patients with transient candidemia, in 1 of 20 otherwise healthy patients with rheumatoid factor, and in 1 patient with a positive cryptococcal latex agglutination test. Three serum specimens from 3 of 32 patients with candidiasis contained rheumatoid factor and gave titers of greater than or equal to 1/8 by the Cand-Tec assay. Detection of serum mannan by enzyme immunoassay was less sensitive but more specific than the Cand-Tec Candida detection system for the diagnosis of invasive candidiasis.