Normalization of quantitative reverse transcription-PCR (Q-RT-PCR) data to appropriate tissue-specific reference genes is an essential part of interpreting the results. This study aimed to determine the most appropriate reference genes for normalizing gene expressions in lymphatic tissue, represented by non-neoplastic lymph nodes and diffuse large B-cell lymphomas, by using 2 statistical software applications, geNorm and NormFinder. In addition, we wanted to validate the usefulness of paraffin-embedded samples for Q-RT-PCR studies by investigating gene expressions of relevant target genes in paired frozen and paraffin-embedded samples. Moreover, we studied the impact of amplicon sizes on the efficiency of Q-RT-PCR in paraffin-embedded tissues. Six putative reference genes were tested for stability of expression in 21 pairs of snap-frozen and formalin-fixed, paraffin-embedded lymph nodes and lymphomas. The genes were ranked according to their suitability as reference genes. According to both statistical approaches, beta-glucoronidase was the single most appropriate reference gene in both snap-frozen and paraffin-embedded samples. TATA box-binding protein gene and Abelson murine leukemia viral oncogene homolog 1 gene were also highly ranked by both programs. In addition, we measured the relative expressions of 7 target genes by Q-RT-PCR, using PCR primer-probes with amplicon sizes up to 105 bases. The correlation coefficient for expression measured in matched frozen and paraffin-embedded samples was 0.93 (P<0.01) after normalization with the appropriate reference genes. Thus, we show that formalin-fixed, paraffin-embedded lymphoid samples are suitable for Q-RT-PCR when using thoroughly validated reference genes.