OBJECTIVES: Colon cancer (CC) is primarily a disease of older individuals, with 65.3% of incident cases in the United States occurring in those aged 65 and above.  There is uncertainty, however, regarding the effectiveness of oxaliplatin in combination adjuvant chemotherapy for prevention of mortality in this population.  Oxaliplatin’s rapid adoption for treatment of stage III colon cancer patients in 2004 provides an opportunity to use calendar time as an instrumental variable (IV) in nonexperimental comparative effectiveness research.  This method makes assumptions different from covariate adjustment and may thus enhance our understanding of treatment effects.  METHODS: We examined patients aged 65+ initiating chemotherapy between 2003-08 using population-based cancer registry data linked with Medicare claims (N=3660).  We derived risk of all-cause mortality from Kaplan-Meier survival curves and estimated risk differences (RD).   We examined IV strength and compared RDs with propensity score (PS)-matched estimates.  RESULTS: Calendar time greatly affected treatment receipt.  The optimal 2-level calendar time instrument achieved a strength of 53.97% compliance by grouping patients treated from January 2003 through September 2004 (n=1449) and those treated from March 2005 through May 2007 (n=1432).   The 1-,2-,3-year Wald IV RDs (95% confidence interval) were -0.05 (-0.09, -0.01), -0.07 (-0.12, -0.01), -0.09 (-0.15, -0.03), respectively. PS-adjusted 1-,2-,3-year RDs were -0.02(-0.04,0), -0.03(-0.06,-0.01), -0.04(-0.08,-0.01). CONCLUSIONS: IV and PS-adjusted models both indicate a survival advantage in this population of older patients treated with oxaliplatin, albeit with different point estimates. As these RDs are based on different assumptions and apply to different populations, the IV analysis strengthens evidence of oxaliplatin's effectiveness in older adults, who bear the greatest burden of colon cancer yet were underrepresented in oxaliplatin clinical trials. In nonexperimental CER of emerging therapies, the potential to use calendar time as an IV should be considered.