We compared the efficacy and safety of gemcitabine, cisplatin, prednisone and thalidomide (GDPT) with standard CHOP regimen (cyclophosphamide, doxorubicin, vincristine, prednisone) for patients with newly diagnosed peripheral T-cell lymphoma (PTCL) in a prospective randomized controlled and open-label clinical trial. Between July 2010 and June 2016, 103 patients were randomly allocated into two groups, of whom 52 were treated with GDPT therapy and 51 with CHOP therapy. The 2-year progression-free survival (PFS) and overall survival (OS) rates were better in the GDPT group than in the CHOP group (57% vs. 35% for 2-year PFS, P = 0·0035; 71% vs 50% for 2-year OS, P = 0·0001). The complete remission rate (CRR) and the overall response rate (ORR) in the GDPT group were higher than in the CHOP group (52% vs. 33%, P = 0·044 for CRR; 67% vs. 49%, P = 0·046 for ORR). Haemocytopenia was the predominant adverse effect, and acute toxicity was moderate, tolerable and well managed in both arms. mRNA expression of ERCC1, RRM1, TUBB3 and TOP2A genes varied among patients but the difference did not reach statistical significance, mainly due to the relatively small sample size. The precise characters of these biomarkers remain to be identified. In conclusion, GDPT is a promising new regimen as potential first-line therapy against PTCL. This study was registered at www.clinicaltrials.gov as #NCT01664975.