Background. This study sought to identify the existence of disparities in the use of newer chemotherapy and to measure the clinical and economic impact of such therapy in SEER-Medicare colorectal cancer (CRC) patients. Methods. Data on stage II, III, IV and unknown stage SEER-Medicare patients diagnosed with CRC between 1998 and 2002 were used. Irinotecan (IR), marketed at the start of the study period, was defined as 'new' therapy. Descriptive statistics were generated and a multivariable logistic regression run to estimate odds of IR receipt in older patients (vs. younger ones), and in nonwhite races (vs. white) within the first two months of chemotherapy initiation. Cox proportional hazards modeling was used to estimate the survival benefit associated with IR use; a descriptive analysis of Medicare payments and a two-part multivariable regression analysis evaluating the association of IR use within a chemotherapy episode with supportive drug payments made by Medicare within that episode also were run.;Results. The final patient sample was composed of 34,863 stage II, III, IV and unknown stage CRC patients. Among chemotherapy users (14,189), use of IR as a first regimen was observed primarily in stage IV patients---approximately 25% of the 3,327 chemotherapy-using stage IV patients initiated therapy with IR. Consequently, all further analyses were applied to stage IV patients. With respect to disparities, older patients and black patients were significantly less likely to receive IR than younger patients and white patients, respectively. The age disparity was modified by gender. With respect to the survival benefit of IR, a clear significant advantage could not be shown. The economic impact of IR was substantial in terms of direct drug costs; IR-users also were significantly more likely to incur costs due to increased use of supportive drugs. Conclusion. The data suggested the existence of age and racial disparities in the receipt of newer chemotherapy as exemplified by IR. Given use of IR, no significant benefit was observed for the agent, although payments for supportive drugs increased. Some limitations to the study existed, including generalizability, and lack of information on certain variables such as patient preferences.