Invasive fungal infection has become a major cause of morbidity and mortality in immunocompromised patients. Rapid identification of pathogenic fungi to species level is critical for disease treatment. A real-time LightCycler assay aiming at rapid detection and species identification of pathogenic fungi from clinical isolates was developed. Template DNAs of different species were amplified and detected in real time by employing SYBR Green fluorescent dye. The target sequences for species-level detection were located between the 18S and 28S rDNA. Seven fungal species encountered frequently in the clinical setting, Candida albicans, Candida glabrata, Candida krusei, Candida parapsilosis, Candida tropicalis, Candida guilliermondii and Cryptococcus neoformans, could be discriminated by species-specific primers and confirmed by melting-curve analyses. The range of linearity was from 1 ng to 1 pg (microl(-1) water) and the sensitivity was 1 pg fungal DNA microl(-1). Identification by this real-time PCR method matched biochemical identification for all 58 clinical strains. Therefore, the method is simple, rapid and sensitive enough for detection and identification of several fungal species.