HIV-associated lymphomas (HALs) have high rates of latent infection by gammaherpesviruses (GHVs). We hypothesized that proteasome inhibition would induce lytic activation of GHVs and inhibit HIV infectivity via preservation of cytidine deaminase APOBEC3G, improving lymphoma control. We tested this oncolytic and antiviral strategy by using bortezomib combined with ifosfamide, carboplatin, and etoposide (ICE) alone or with rituximab (ICE/R) in relapsed/refractory HAL. A 3+3 dose-escalation design was used with a 7-day lead-in period of single-agent bortezomib. Bortezomib was administered intravenously on days 1 and 8 of each cycle at 1 of 4 dose levels: 0.7, 1.0, 1.3, or 1.5 mg/m2 ICE began day 8 of cycle 1 and day 1 of subsequent cycles. Rituximab was included on day 1 of cycles 2 to 6 for CD20+ lymphomas. Twenty-three patients were enrolled. The maximum tolerated dose of bortezomib was not reached. Grade 4 toxicities attributable to bortezomib were limited to myelosuppression. Responses occurred in 17 (77%) of 22 patients receiving any protocol therapy. The 1-year overall survival was 57%. After bortezomib alone, both patients with Kaposi sarcoma herpesvirus (KSHV)-positive lymphoma had more than a 1-log increase in KSHV viral load. In 12 patients with Epstein-Barr virus (EBV)-positive lymphoma, median values of EBV viral load increased. Undetectable HIV viremia at baseline in the majority of patients limited evaluation of HIV inhibition. APOBEC3G levels increased in 75% of evaluable patients. Bortezomib combined with ICE/R in patients with relapsed/refractory HAL is feasible with response and survival comparing favorably against previously reported second-line therapies. Changes in GHV viral loads and APOBEC3G levels trended as hypothesized. This trial was registered at www.clinicaltrials.gov as #NCT00598169.