Many inner-city residents with progressive renal disease do not receive specialist care from a nephrologist, and often arrive in the emergency room manifesting life-threatening uremic symptoms requiring emergency rescue dialysis. We examined the relationship between the quality of medical care received during progression to end-stage renal disease, and the clinical condition and morbidity at initiation of renal replacement therapy. During a 5-year period (January 1990 to December 1994), we prospectively studied 139 consecutive inner-city residents with a confirmed diagnosis of chronic renal failure who were starting uremia therapy. At onset of study, subjects were sorted into one of three groups depending on the extent of medical care received during the 3 years immediately preceding initiation of hemodialysis: nephrologist, nonnephrologist (physician), or no medical care. Information obtained from each subject included length of hospital stay during the admission for initiation of dialysis therapy and the type of hemodialysis vascular access used for first dialysis treatment (permanent v temporary). Predialysis blood urea nitrogen concentration, serum creatinine concentration, serum albumin concentration, and serum bicarbonate concentration were measured once immediately before the first dialysis. The 139 study subjects (62 men and 77 women) comprised 116 blacks (83%), 15 Hispanics (11%), and eight whites (6%), and had a mean age of 56 +/- 15 years (+/-SD). Only 59 (43%) subjects received prior specialist nephrologist care, and their mean length of hospital stay (12 +/- 23 days) was shorter than that of subjects who received nonnephrologist care (n = 63 [45%]; 25 +/- 21 days) or those who received no prior medical care (n = 17 [12%]; 29 +/- 23 days) (P = 0.002). Temporary hemodialysis vascular access was used for the first dialysis in all 17 (100%) of the subjects with no prior medical care, in 56 (89%) of the 63 subjects who received prior care from a nonnephrologist, and in 21 (36%) of the 59 subjects who received prior care from a nephrologist (P = 0.0001). Subjects who received prior care from a nephrologist had a lower mean serum creatinine concentration at initiation of dialysis (11 +/- 4.4 mg/dL) than did either the subjects who received nonnephrologist care (13 +/- 5.4 mg/dL) or no medical care (16 +/- 5.7 mg/dL) (P = 0.003). In addition, subjects who received prior care from a nephrologist had less severe metabolic acidosis than the subjects in the other two groups (P = 0.04). We infer that initiation of uremia therapy is delayed in inner-city residents with progressive renal failure who do not receive specialist nephrologic care, and that as a consequence these patients suffer excess short-term morbidity.