Canagliflozin slows the progression of chronic kidney disease in patients with type 2 diabetes and induces a reversible acute drop in estimated glomerular filtration rate (eGFR), believed to be a hemodynamic effect. Predictors of the initial drop and its association with long-term eGFR trajectories and safety outcomes are unknown. To assess this, we performed a post-hoc analysis of 4289 participants in the CREDENCE trial with type 2 diabetes and chronic kidney disease equally split into treatment and placebo groups who had eGFR measured at both baseline and week three. The eGFR was categorized at week three as greater than a 10% decline; between 0 and 10% decline; and no decline. Long-term eGFR trajectories and safety outcomes were estimated in each category of acute eGFR change by linear mixed effects models and Cox regression after adjustment for baseline characteristics and medications use. Significantly more participants in the canagliflozin (45%) compared to the placebo (21%) group experienced an acute drop in eGFR over 10%. An over 30% drop occurred infrequently (4% of participants with canagliflozin and 2% with placebo). The odds ratio for a drop in eGFR over 10% with canagliflozin compared to placebo was significant at 3.03 (95% confidence interval 2.65, 3.47). Following the initial drop in eGFR, multivariable adjusted long-term eGFR trajectories, as well as overall and kidney safety profiles, in those treated with canagliflozin were similar across eGFR decline categories. Thus, although acute drops in eGFR over 10% occurred in nearly half of all participants following initiation of canagliflozin, the clinical benefit of canagliflozin was observed regardless. Additionally, safety outcomes were similar among subgroups of acute eGFR drop.