Since non-anthracycline-based chemotherapy with L-asparaginase has improved survival outcomes in patients with extranodal natural killer/T-cell lymphoma (ENKTL), the incidence of central nerve system (CNS) relapse can be different when compared with previous reports. In this research, we sought to identify the incidence of and predictors for CNS relapse and to evaluate the necessity of CNS prophylaxis with intermediate-dose MTX (ID-MTX). The records of 399 patients in the training cohort and 253 patients in the validation cohort with ENKTL who received non-anthracycline-based chemotherapy were reviewed. Patients were divided into two groups according to whether the chemotherapy regimen included ID-MTX above 2 g/m2. A new CNS-PINK model was developed using one-point powerful predictors of CNS relapse [the prognostic index of natural killer lymphoma (PINK); HR: 2.908; P = .030 and extranodal involvement (≥ 2); HR: 4.161; P = .001] and was calculated as a sum of scores. The high-risk group of CNS-PINK was defined as 2 points. The cumulative incidence of CNS relapse was different between the CNS-PINK risk groups in the training (P < .001) and validation cohort (P = .038). Patients in the high-risk CNS-PINK group who received SMILE or SMILE-like regimens with ID-MTX (S-ID-MTX) displayed a lower incidence rate of CNS relapse than did those who received other regimens without ID-MTX in the training cohort (P = .029). The CNS-PINK was demonstrated its strong predictability of CNS relapse in ENKTL patients. The effect of S-ID-MTX for preventing CNS events in high-risk CNS-PINK patients should be verified by future studies.