The recent WHO classification of haematological malignancies includes detection of genetic abnormalities with rognostic significance. Consequently, an increasing number of specific real-time quantitative reverse transcription polymerase chain reaction (Q-RT-PCR) based assays are in clinical research use and needs qualit control for accuracy and precision. Especially the identification of experimental variations and statistical analysis has recently created discussions. The standard analytical technique is to use the Delta-Delta-Ct method. Although this method accounts for sample specific variations such as RNA purification, it does not account for other experimental effects as variations in cDNA synthesis, amplification efficiency and assay variations. To obtain an assessment of the accuracy and precision of the assays a novel approach for the statistical analysis of Q-RT-PCR has been developed based on a linear mixed effects model for factorial designs. The model consists of an analysis of variance where the variation of each fixed effect of interest and identified experimental and biological nuisance variations are split. Hereby it accounts for varying efficiency, inhomogeneous variance, repeated measures correlation and experimental variations. The modelling approach has been used to conduct fold change analysis on microRNA (miRNA) expressions in Diffuse Large B-cell Lymphoma (DLBCL). In particular it was demonstrated that results obtained from global miR expression arrays (Exiqon) could be reproduced and validated by Q-RT-PCR.