BACKGROUND: (blind field). METHODS: Records for patients treated for newly diagnosed or recurrent glioblastoma were analyzed. All patients had undergone surgical resection and tumor specimens at time of surgery were available for culture in a 3-dimensional matrix assay and observed for growth and invasion. Drug effects on mean invasion and growth were expressed as a ratio relative to control conditions. Length of survival was compared between temozolomide treated patients whose screening results had predicted a positive or negative response to temozolomide. The MGMT status of a subgroup of these patients was analyzed and correlated with the response of tumor tissue in the assay to temozolomide. RESULTS: Fifty-eight patients with glioblastoma were assessed. Each patient's tumor displayed a unique invasion and response profile. We looked in particular at the correlation between the outcome of a patient with glioblastoma treated with temozolomide and the response of that patient's tumor tissue to temozolomide in the 3-dimensional assay. Mean survival time for patients whose tumors were not significantly sensitive to temozolomide in the assay was 181.7 +/- 43 days. Mean survival time for patients whose tumors were significantly sensitive to temozolomide in the assay was 290.0 +/- 33 days. Twelve patients underwent MGMT testing. In 10 of the 12 patients there was a correlation between tumor response in the assay and MGMT status. CONCLUSIONS: The 3-dimensional assay may help predict glioblastoma patients who will show a treatment response to temozolomide. There appears to be a positive correlation between the response profiles in the assay to the MGMT status of the patient's tumor. SECONDARY CATEGORY: n/a.