Background and objectives The secular trend toward dialysis initiation at progressively higher levels of eGFR is not well understood. This study compared temporal trends in eGFR at dialysis initiation within versus outside the Department of Veterans Affairs (VA)-the largest non-fee-for-service health system in the United States.Design, setting, participants, & measurements The study used linked data from the US Renal Data System, VA, and Medicare to compare temporal trends in eGFR at dialysis initiation between 2000 and 2009 (n=971,543). Veterans who initiated dialysis within the VA were compared with three groups who initiated dialysis outside the VA: (1) veterans whose dialysis was paid for by the VA, (2) veterans whose dialysis was not paid for by the VA, and (3) nonveterans. Logistic regression was used to estimate average predicted probabilities of dialysis initiation at an eGFR$10 ml/min per 1.73 m 2 .Results The adjusted probability of starting dialysis at an eGFR$10 ml/min per 1.73 m 2 increased over time for all groups but was lower for veterans who started dialysis within the VA (0.31; 95% confidence interval [95% CI], 0.30 to 0.32) than for those starting outside the VA, including veterans whose dialysis was (0.36; 95% CI, 0.35 to 0.38) and was not (0.40; 95% CI, 0.40 to 0.40) paid for by the VA and nonveterans (0.39; 95% CI, 0.39 to 0.39). Differences in eGFR at initiation within versus outside the VA were most pronounced among older patients (P for interaction ,0.001) and those with a higher risk of 1-year mortality (P for interaction ,0.001).Conclusions Temporal trends in eGFR at dialysis initiation within the VA mirrored those in the wider United States dialysis population, but eGFR at initiation was consistently lowest among those who initiated within the VA. Differences in eGFR at initiation within versus outside the VA were especially pronounced in older patients and those with higher 1-year mortality risk.