In chronic kidney disease, glycated albumin and fructosamine have been postulated to be better biomarkers of glycemic control than HbA1c. We evaluated the accuracy, variability, and covariate bias of three biomarkers (HbA1c, glycated albumin, and fructosamine) compared with continuous glucose monitoring (CGM)-derived measurement of glycemia across estimated glomerular filtration rate (eGFR) in type 2 diabetes.Prospective cohort study of 104 participants with type 2 diabetes, 80 with eGFR <60 mL/min/1.73 m2 (not treated with dialysis) and 24 frequency-matched control subjects with eGFR ≥60 mL/min/1.73 m2. Participants wore a blinded CGM for two 6-day periods separated by 2 weeks, with blood and urine collected at the end of each CGM period. HbA1c, glycated albumin, and fructosamine were measured by high-performance liquid chromatographic, enzymatic, and colorimetric nitroblue tetrozolium methods, respectively.Within-person biomarker values were strongly correlated between the two CGM periods (r = 0.92-0.95), although no marker fully captured the within-person variability of mean CGM glucose. All markers were similarly correlated with mean CGM glucose (r = 0.71-77). Compared with mean CGM glucose, glycated albumin and fructosamine were significantly biased by age, BMI, serum iron concentration, transferrin saturation, and albuminuria; HbA1c was underestimated in those with albuminuria.Glycated albumin and fructosamine were not less variable than HbA1c at a given mean CGM glucose level, with several additional sources of bias. These results support measuring HbA1c to monitor trends in glycemia among patients with eGFR <60 mL/min/1.73 m2. Direct measurements of glucose are necessary to capture short-term variability.