Puccinia horiana Henn. is a quarantine organism and one of the most important fungal pathogens of Chrysanthemum x morifolium cultivars grown for cut flower or potted plant production (florist's chrysanthemum) in several regions of the world. Highly specific primer pairs were identified for conventional, nested, and real-time PCR detection of P. horiana based on the specific and sensitive PCR amplification of selected regions in the internal transcribed spacers (ITS1 and ITS2) of the nuclear ribosomal DNA (rDNA). Using these different PCR versions, 10 pg, 10 fg, and 5 fg genomic DNA could be detected, respectively. When using cloned target DNA as template, the detection limits were 5000, 50, and 5 target copies, respectively. These detection limits were not affected by a background of chrysanthemum plant DNA. The DNA extraction method was optimized to maximize the recoverability of the pathogen from infected plant tissue. A CTAB extraction protocol or a selection of commercial DNA extraction methods allowed the use of 10 ng total (plant+pathogen) DNA without interference of PCR inhibitors. Due to the specificity of the primers, SYBR Green I technology enabled reliable real time PCR signal detection. However, an efficient TaqMan probe is available. The lowest proportion of infected plant material that could still be detected when mixed with healthy plant material was 0.001%. The real-time PCR assay could detect as few as eight pure P. horiana basidiospores, demonstrating the potential of the technique for aerial detection of the pathogen. The amount of P. horiana DNA in plant tissue was determined at various time points after basidiospore inoculation. Using the real-time PCR protocol, it was possible to detect the pathogen immediately after the inoculation period, even though the accumulation of pathogen DNA was most pronounced near the end of the latent period. The detection system proved to be accurate and sensitive and could help not only in pathogen diagnosis but also in pathogen monitoring and disease forecasting systems.