OBJECTIVE The tolerance of escalating targeted doses of carboplatin combined with ifosfamide (IFOS)/etoposide (VP-16) (ICE) was assessed in children with recurrent solid tumors. METHODS To reduce interpatient variability in carboplatin systemic exposure, 45 children were treated with doses individualized to a target area under the serum concentration versus time curve (AUC) based on renal function, using technetium 99-diethyl-enetriamine pentaacetic acid (99mTc-DTPA) clearance to estimate glomerular filtration rate (GFR). Cohorts of at least three patients received carboplatin at an initial target AUC of 2 mg/mL x min, with escalations of 1 mg/mL x min in subsequent cohorts. Courses consisted of carboplatin on day 1 followed by IFOS 2 g/m2 plus VP-16 100 mg/m2 on days 2 and 3. Patients received at least two courses, with a maximum of eight courses possible in the absence of progressive disease. When only moderate toxicity occurred after escalation to 5 mg/mL x min, a third dose of IFOS plus VP-16 was added. After three patients were treated at this level, carboplatin escalation proceeded. RESULTS Neutropenia and thrombocytopenia were the dominant toxicities in the 43 assessable patients. At the target AUC of 8 mg/mL x min, 13 of 20 cycles were associated with febrile neutropenia. For phase II trials, we recommend a carboplatin target AUC of 6 mg/mL x min with three doses of IFOS and VP-16 for patients with prior craniospinal irradiation or high-dose cisplatin (CDDP)/VP-16, or 7 mg/mL x min for patients without such histories. There were two complete responses (CRs), 13 partial responses (PRs), and 17 objective responses (ORs). CONCLUSIONS The ICE regimen shows promising activity in pediatric solid tumors. The clear relationship between hematologic toxicity and carboplatin systemic exposure supports the use of targeted dosing in further trials of ICE chemotherapy.