Nucleic acid amplification of clinical specimens with low target concentration has variable sensitivity. We examined whether testing multiple aliquots of extracted DNA increased the sensitivity and reproducibility of Chlamydia pneumoniae detection by PCR. Nested and non-nested C. pneumoniae PCR assays were compared using 10 replicates of 16 serial dilutions of C. pneumoniae ATCC VR-1310. The proportion positive versus the C. pneumoniae concentration was modeled by probit regression analysis. To validate the model, 10 replicates of 26 previously positive patient specimens of peripheral blood mononuclear cells (PBMC), sputum, or nasopharyngeal swabs (NPS) were tested. The proportion of replicates that were positive varied with the concentration of C. pneumoniae in the sample. At concentrations above 5 infection-forming units (IFU)/ml, both nested and non-nested PCR assay sensitivities were 90% or greater. The nested PCR was more sensitive (median detection, 0.35 versus 0.61 IFU/ml; relative median detection, 0.58; 95% confidence interval, 0.31 to 0.99; P = 0.04). In clinical specimens, replicate PCR detected 15 of 26 (nested) versus 1 of 26 (non-nested, P < 0.001). For PBMC specimens, testing 1, 3, or 5 replicates detected 3, 5, or 9 of 10 positive specimens, respectively. Median C. pneumoniae concentrations were estimated at 0.07 IFU/ml for PBMC and at <0.03 IFU/ml for NPS specimens. We conclude that performing 5 or 10 replicates considerably increased the sensitivity and reproducibility of C. pneumoniae PCR and enabled quantitation for clinical specimens. Due to sampling variability, PCR tests done without replication may miss a large proportion of positive specimens, particularly for specimens with small amounts of target C. pneumoniae DNA present.