Clinical trials to assess the treatment of diabetic and other forms of polyneuropathy are becoming increasingly common. Nerve conduction studies, quantitative sensory testing, and autonomic function testing are often used in these trials. This article reviews the sensitivity and reproducibility of these measures to detect change in peripheral nerve function during long-term trials. The attributes of nerve conduction studies that are likely to be most useful are summated or averaged sensory nerve action potential amplitudes and averaged motor nerve conduction velocities. Summated or averaged compound muscle action potential amplitude and mean F-wave latencies are also highly informative. Vibratory detection thresholds are sensitive, specific, and highly reproducible for assessment of large myelinated sensory fibers, with cooling and warming detection thresholds also having good sensitivity for small myelinated sensory fibers. Although less well validated for longitudinal trials, visual analogue scale scoring of heat pain provides assessment of unmyelinated sensory fibers. Heart rate variation to deep breathing, Valsalva, or standing are useful to assess cardiac autonomic function. Based on these data that are reviewed and consistent with the conclusions of previous consensus conferences, a combination of these studies is recommended.