Our aim was to determine several indexes of reproducibility for strength measurements with a hand-held pull-gauge dynamometer (MFB50K) in cancer patients. Two independent testers performed repeated measurements of maximal isometric elbow and knee strength. The measurements were gathered in a convenience sample of 40 patients (27 men, age 20-72 years) with various types and stages of cancer. These patients were recruited from the medical oncology clinic of a university hospital. The Pearson product moment correlation coefficient, the intraclass correlation coefficient (ICC), the standard error of measurement (SEM) and its associated 95% confidence interval, the smallest detectable difference (SDD) and the limits of agreement were determined as outcome measures for reproducibility. The results for the Pearson product moment correlation and the ICC were 0.90 for elbow and 0.96 for knee extension strength. The results in terms of the SDDs were 29.4 newton (N) for elbow and 54.8 N for knee strength. The limits of agreement for elbow strength were -21.7 N and 38.5 N (mean 8.4 N). For knee strength the limits of agreement were -49.8 N and 59.8 N (mean 5 N). The Pearson product moment correlation and the ICC suggested good reproducibility. However, the more informative indices of reproducibility, i.e. the SEM and the SDD, showed relatively large measurement error between the testers. Therefore, the current use of the MFB50K for the measurement of muscle strength is not supported in cancer patients.