Effective treatment of diabetic sensorimotor polyneuropathy remains a challenge. To assess the efficacy and safety of α-lipoic acid (ALA) over 20 weeks, we conducted a multicenter randomized withdrawal open-label study, in which 45 patients with type 2 diabetes and symptomatic polyneuropathy were initially treated with ALA (600 mg tid) for 4 weeks (phase 1). Subsequently, responders were randomized to receive ALA (600 mg qd; n = 16) or to ALA withdrawal (n = 17) for 16 weeks (phase 2). During phase 1, the Total Symptom Score (TSS) decreased from 8.9 ± 1.8 points to 3.46 ± 2.0 points. During phase 2, TSS improved from 3.7 ± 1.9 points to 2.5 ± 2.5 points in the ALA treated group (p < 0.05) and remained unchanged in the ALA withdrawal group. The use of analgesic rescue medication was higher in the ALA withdrawal group than ALA treated group (p < 0.05). In conclusion, in type 2 diabetic patients with symptomatic polyneuropathy who responded to initial 4-week high-dose (600 mg tid) administration of ALA, subsequent treatment with ALA (600 mg qd) over 16 weeks improved neuropathic symptoms, whereas ALA withdrawal was associated with a higher use of rescue analgesic drugs. This trial is registered with ClinicalTrials.gov Identifier: NCT02439879.