Recent clinical drug trials designed to test the effect on established mild diabetic neuropathy have in general been disappointing. These findings may in part be due to a failure of tested drugs to reverse neuropathy (they may merely halt its progression) and to insufficient durations of the trials. To aid the design of future studies, we examined the progression rates of quantitative sensory tests, autonomic functions, and sensory and motor nerve electrophysiology in 182 patients designed to placebo treatment in an 18-month multicenter ARI-trial. Clinically meaningful deteriorations were demonstrated in the vibratory perception threshold in the toe and the Valsalva ratio. The greatest deterioration rate in electrophysiologic measures was found in peroneal F-wave latency and in sensory nerve conduction velocities in the upper limb, but none of these reached the threshold of clinically meaningful change. Assuming that drug efficacy will be based on the deterioration rates in placebo patients alone, the present data suggest a minimum of 250 patients treated for at least 2 years to achieve convincing efficacy.