Human T‐cell leukemia virus type‐1 (HTLV‐1) proviral load (VL) is an important determinant of viral pathogenesis and malignant evolution. Although VL has been quantified by in‐house real‐time quantifiable polymerase chain reaction (qPCR) technology, little is known about the harmonization among different VL assay systems. We evaluated intra‐ and inter‐laboratory variability of VL measured at six laboratories using the same DNA samples seropositive for HTLV‐1 in a two‐step manner. The first study measured 60 samples by original in‐house assays, finding that the median intra‐ and inter‐laboratory coefficient of variation (CV) was 44.9% (range, 25.4–71.8%) and 59.9% (34.2–93.4%), respectively. The inter‐laboratory correlation coefficients ranged from 0.760 to 0.875, indicating that VL were measured with good precision in each laboratory, but inter‐laboratory regression slopes differed from 0.399 to 2.206, indicating that VL were measured with a wide variation between laboratories. To examine the effect of standardization of reference materials (RM) on the VL variability, we performed a second study using only 20 samples by substituting RM for plasmid including the HTLV‐1 pX region. The median inter‐laboratory CV for raw pX copy number was reduced significantly from 66.9% to 35.7%, whereas the median CV for the internal control remained almost unchanged, resulting in no improvement in inter‐laboratory CV for VL. This indicates that each in‐house assay system worked well with good precision, but standardizing RM alone was insufficient for harmonization. The relevant choice of not only RM, but also internal control genes for data normalization is expected to be realistic to standardize HTLV‐1 VL measurement. (Cancer Sci 2010; 101: 2361–2367)