In 3 large cohorts (total of 787 patients), clinical vibration impairment (CVI) of the great toe using a tuning fork was compared with quantitative vibration threshold (QVT). Using a stepwise multivariate analysis, we assessed demographic and anthropomorphic patient characteristics associated with the difference between CVI and QVT for the various cohorts and the chosen QVT ranges of percentile abnormality. We also compared CVI or QVT abnormality with a composite score of nerve conduction abnormality to confirm that QVT is a valid measure of severity of neuropathy. Highly significant associations between CVI and QVT were found in all 3 cohorts studied, regardless of the chosen QVT percentile level of abnormality. However, in the 2 cohorts evaluated by many different physicians, CVI overestimated QVT much more often than underestimated it. The discordance between CVI and QVT in all QVT abnormality percentile levels was associated with age, height and body surface area (BSA) in 1 cohort, with age and BSA in another cohort, and with age in the third cohort. In the third cohort, the correlation between QVT and the composite score of nerve conduction abnormality was significantly higher than the correlation between CVI and the composite score. Using a tuning fork, neuromuscular physicians overestimate vibration sensation loss more often than when QVT testing is done, which employs quantitative stimuli, a broad range of stimulus magnitudes, null stimuli, validated algorithms of testing and validated reference values. To improve assessment of vibration sensation, physicians should take into account age, height and weight (or body surface area) when judging vibration abnormalities. Applying some useful approaches to quantitative sensory testing might improve the accuracy of clinical sensory testing.