BACKGROUND: Hand-held dynamometry is a portable and inexpensive method to quantify muscle strength. To determine if muscle strength has changed, an examiner must know what part of the difference between a patient's pre-treatment and post-treatment measurements is attributable to real change, and what part is due to measurement error. This study aimed to determine the relative and absolute reliability of intra and inter-observer strength measurements with a hand-held dynamometer (HHD). METHODS: Two observers performed maximum voluntary peak torque measurements (MVPT) for isometric knee extension in 24 patients with haematological malignancies. For each patient, the measurements were carried out on the same day. The main outcome measures were the intraclass correlation coefficient (ICC +/- 95%CI), the standard error of measurement (SEM), the smallest detectable difference (SDD), the relative values as % of the grand mean of the SEM and SDD, and the limits of agreement for the intra- and inter-observer ;3 repetition average; and the ;highest value of 3 MVPT; knee extension strength measures. RESULTS: The intra-observer ICCs were 0.94 for the average of 3 MVPT (95%CI: 0.86-0.97) and 0.86 for the highest value of 3 MVPT (95%CI: 0.71-0.94). The ICCs for the inter-observer measurements were 0.89 for the average of 3 MVPT (95%CI: 0.75-0.95) and 0.77 for the highest value of 3 MVPT (95%CI: 0.54-0.90). The SEMs for the intra-observer measurements were 6.22 Nm (3.98% of the grand mean (GM) and 9.83 Nm (5.88% of GM). For the inter-observer measurements, the SEMs were 9.65 Nm (6.65% of GM) and 11.41 Nm (6.73% of GM). The SDDs for the generated parameters varied from 17.23 Nm (11.04% of GM) to 27.26 Nm (17.09% of GM) for intra-observer measurements, and 26.76 Nm (16.77% of GM) to 31.62 Nm (18.66% of GM) for inter-observer measurements, with similar results for the limits of agreement. CONCLUSION: The results indicate that there is acceptable relative reliability for evaluating knee strength with a HHD, while the measurement error observed was modest. The HHD may be useful in detecting changes in knee extension strength at the individual patient level.