Children with relapse of T-cell acute lymphoblastic leukemia (T-ALL) or lymphoblastic lymphoma (T-LBL) have a dismal prognosis, largely due to difficulty attaining second remission. We hypothesized that adding etoposide and cyclophosphamide to the nucleoside analog nelarabine could improve response rates over single-agent nelarabine for relapsed T-ALL and T-LBL. This phase I dose-escalation trial's primary objective was to evaluate the dose and safety of nelarabine given in combination with etoposide at 100 mg/m(2)/day and cyclophosphamide at 330-400 mg/m(2)/day, each for 5 consecutive days in children with either T-ALL (13 patients) or T-LBL (10 patients). Twenty-three patients were treated at three dose levels; 21 were evaluable for dose-limiting toxicities (DLT) and response. The recommended phase II doses (RP2D) for this regimen, when given daily x5 every 3 weeks, were nelarabine 650 mg/m(2)/day, etoposide 100 mg/m(2)/day, and cyclophosphamide 400 mg/m(2)/day. DLTs included peripheral motor and sensory neuropathies. An expansion cohort to evaluate responses at the RP2D was terminated early due to slow accrual. The overall best response rate was 38% (8/21), with 33% (4/12) responses in the T-ALL cohort and 44% (4/9) responses in the T-LBL cohort. These response rates are comparable to those seen with single-agent nelarabine in this setting. These data suggest that the addition of cyclophosphamide and etoposide to nelarabine does not increase the incidence of neurologic toxicities or the response rate beyond that obtained with single-agent nelarabine in children with first relapse of T-ALL and T-LBL.