The aim of the present study was to prospectively evaluate the recurrence and survival outcome of primary central nervous system lymphoma (PCNSL) with intraocular involvement. For this purpose, a prospective cohort of 103 pathologically confirmed patients with PCNSL was enrolled in this study. Ophthalmologic examinations were performed both at diagnosis and during follow-up. The patients with PCNSL with suspected intraocular involvement underwent vitrectomy for confirmation. Patients who presented with intraocular involvement either at diagnosis or during disease progression were allocated to the intraocular lymphoma (IOL) group. All patients with PCNSL received systemic methotrexate (MTX)-based chemotherapy with or without radiotherapy. MTX intravitreal injection combined with systemic MTX-based chemotherapy was recommended once ocular lesions were confirmed. Recurrent intraocular and central nervous system (CNS) events, progression-free survival (PFS) and overall survival (OS) outcomes were analyzed. The findings of this study revealed that 21 patients with PCNSL exhibited intraocular involvement. One patient with IOL presented with isolated ocular lymphoma at the initial diagnosis, and the others presented with ocular involvement along with CNS invasion during the course of the disease. A total of 14 patients received systemic MTX-based chemotherapy prior to the diagnosis of IOL. The recurrence rates in patients with or without intraocular involvement were 71.4 and 46.3%, respectively (P=0.04) and the relapse sites in the patients with IOL included the brain (3 patients), eyes (6 patients), and both brain and eyes (6 cases). The median PFS was 13 months in the IOL group and 19 months in the patients without intraocular involvement (non-IOL) (P=0.019). The median OS was 51 months vs. 56 months, respectively (P=0.312). There was no significant difference in the 2-year PFS and OS rates between the 2 groups (23.8% vs. 23.2%, P=0.951; and 61.9% vs. 41.4%, P=0.093, respectively). On the whole, the findings of this study suggest that patients with IOL have a high risk of relapse and a poor PFS compared to patients without IOL, but a similar OS.