Whether different methods of quantitating blood pressure (BP) in late chronic kidney disease better mimic pathophysiological processes and clinical outcomes remains unclear. In a retrospective study, we determined the association of BP with end-stage kidney disease (ESKD) and all-cause mortality with BP modeled at baseline versus longitudinally with time-varying Cox models as (1) current (most recent) clinic visit, (2) lag (visit immediately preceding the current), (3) cumulative (average of previous measurements), and (4) change from baseline to the most recent. Among 1203 (6913 visits) study patients, the mean age and baseline estimated glomerular filtration rate were 66 and 18 mL·min−1·1.73 m−2), and 40% were female. Patients had a mean of 6.7 BP measurements, 540 (44.8%) reached ESKD, and 141 (11.7%) died. For systolic BP >160, current (hazard ratio [HR], 1.67), cumulative (HR, 1.58), and a rise to >160 from baseline 120 to 160 (HR, 1.60) were associated with ESKD. Similarly, diastolic BP >85 was associated with ESKD when modeled as current (HR, 1.47), lag (HR, 1.63), cumulative (HR, 2.15), or change from baseline (rise to >85 from a baseline of 60–85; HR, 1.62). Both low SBP (<120), when modeled as current (HR, 1.59), cumulative exposure (HR, 1.76), persistently <120 (HR, 2.28), and high SBP (>140), when modeled as cumulative exposure, were associated with all-cause mortality. For diastolic BP, only cumulative >85 was significantly associated with mortality (HR, 2.75). Thus, in late-stage chronic kidney disease, persistently high or rises in systolic BP or diastolic BP are associated with risk of ESKD, whereas baseline BP measures did not convey information on risk.