Small-fiber neuropathy (SFN) is diagnosed on the basis of clinical features and specialized tests of small-fiber function because standard nerve conduction studies are normal. Thus, the objective of this study was to determine the value of deep tendon reflex (DTR) latency measurement in the diagnosis of SFN in patients with preserved DTR on clinical examination. We prospectively examined electromyographic reflexes from the biceps brachii [biceps brachii reflex (BR)], patellar [patellar reflex (PR)], and ankle [ankle reflex (AR)] using a manually operated electronic reflex hammer attached to electromyography machine and recorded by means of surface electrodes in 18 patients with SFN and 38 controls. Intra- and inter-evaluator reliability was good (intraclass correlation coefficient: 0.80-0.91, p < 0.01). In controls, the latencies at all sites were correlated to the height (R= 0.6, p < 0.01). Compared with controls, in patients with SFN, the mean latency in milliseconds was prolonged at all sites (BR: 12.8 +/- 1.6 vs. 8.9 +/- 1.9, p < 0.01; PR: 23.0 +/- 5.8 vs. 17.4 +/- 2.4, p < 0.01; and AR: 34.5 +/- 4.8 vs. 30.0 +/- 2.4, p < 0.01). The sensitivity [61.1% (95% CI: 51-94.9)] and specificity [92% (95% CI: 73-97.3)] of BR latency were roughly equal to those of PR and AR. We conclude that DTR latencies were significantly abnormal in the majority of the patients with SFN, suggestive of subclinical involvement of large myelinated fibers. DTR latency measurement is a reproducible, valuable, sensitive tool in the evaluation of mild subclinical involvement of large fibers.