Discontinuation of life-sustaining dialysis therapy led to death in 282 of 5,208 patients who started therapy for end-stage renal disease (ESRD) in Michigan during 1980-1985 with a follow-up through 1986. Based on life table estimates at 60 months after initiation of therapy, 9.4% of patients overall died due to termination of dialysis, 11% of females versus 8% of males (p = 0.02), 0.1-3.4% for ages less than or equal to 49 years versus 56% for greater than 80 years, 12% for white versus 4% for black patients (p less than 0.001) and 16% for diabetic ESRD patients (higher than any other group, p less than 0.05). The Cox regression model confirms these significant findings for race, diabetes and age, and reveals a significant 60% increase in overall withdrawals for the years 1980-1985 (1.10/year, p less than 0.02). A separate analysis of discontinuation of dialysis as the percentage of all 2,564 dialysis deaths in prevalence cases for 1980-1984 revealed an overall ratio of 8.9% with a significant difference for ages less than or equal to 64 versus greater than or equal to 65 (p less than 0.001), race (p less than 0.001) but not for prior transplant failure or continuous ambulatory peritoneal dialysis therapy. Whereas the results for age and diabetes were expected, the significant increase of dialysis withdrawal over time and the racial difference are unexplained by information available at the Michigan Kidney Registry and indicate the need for exploration by further studies.