Circulating miRNAs are intensively evaluated as promising blood-based biomarkers. This growing interest in developing assays for circulating miRNAs necessitates careful consideration of the effects of preanalytical and analytical parameters on the isolation, stability, and quantification of circulating miRNAs. By using quantitative stem-loop RT-PCR, we compared the relative efficiencies of four miRNA isolation systems and different storage conditions. The effect of the data normalization procedure on the quantification of circulating miRNA levels in plasma from 30 healthy individuals and 30 patients with non–small cell lung carcinoma was estimated by measuring endogenous hsa-miR-21 and hsa-miR-16 and exogenous cel-miR-39 that was spiked in all samples at the same concentration. Silica column–based RNA extraction methods are more effective and reliable with respect to TRIzol LS. Endogenous circulating miRNA levels are unstable when plasma is stored at 4°C, and samples should be kept at −70°C, where the extracted miRNAs remain stable for up to 1 year. When normalization is based on combined endogenous and exogenous control miRNAs, differences in miRNA recovery and differences in cDNA synthesis between samples are compensated. Using this normalization procedure and hsa-miR-21 as a biomarker, we could clearly discriminate healthy individuals from patients with cancer. Experimental handling and the use of exogenous and endogenous controls for normalization are critical for the reliable quantification of circulating miRNA levels in plasma.