To determine whether neuroblastomas acquire a sustained drug-resistant phenotype from exposure to chemotherapeutic agents given to patients in vivo, we studied neuroblastoma cell lines established at different points of therapy: six at diagnosis before therapy (DX), six at progressive disease during induction therapy (PD-Ind), and five at relapse after intensive chemoradiotherapy and bone marrow transplantation (PD-BMT). Cells were maintained in the absence of drug selective pressure. Dose-response curves of melphalan, cisplatin, carboplatin, doxorubicin, and etoposide for the cell line panel were determined by measuring cytotoxicity with a 96-well-plate digital imaging microscopy (DIMSCAN) microassay. Drug resistance of cell lines progressively increased with the intensity of therapy delivered in vivo. The greatest resistance was seen in PD-BMT cell lines: IC90 values in PD-BMT cell lines were higher than clinically achievable drug levels by 1-37 times for melphalan, 1-9 times for carboplatin, 25-78 times for cisplatin, 6-719 times for doxorubicin, and 3-52 times for etoposide. Genomic amplification of MYCN did not correlate with resistance. Cross-resistance by Pearson correlation (r > or = 0.6) was observed between: (a) cisplatin + doxorubicin; (b) carboplatin + cisplatin, etoposide, or melphalan; (c) etoposide + cisplatin, melphalan, or doxorubicin. These data indicate that during therapy, neuroblastomas can acquire resistance to cytotoxic drugs because of the population expansion of tumor cells possessing stable genetic or epigenetic alterations that confer resistance.