AQP5 has been shown to be upregulated in kidney biopsies from patients with diabetic nephropathy.Here we investigate whether urinary baseline AQP5 is independently associated with eGFR decline in patients with type 2 diabetes and nephropathy. Baseline urine samples (n=997) from patients with type 2 diabetes and nephropathy recruited in the Sun-MACRO randomized placebo controlled double blind clinical trial were used for AQP5 measurement, using human AQP5-specific enzyme linked immunosorbent kits. Pearson correlation and multiple linear regression between AQP5 with eGFR slope (calculated by ≥3 serum creatinine during followup) was performed, and association with fast renal function decline, defined as eGFR slope less than 3.0 mL/min/1.73m 2 /year, was determined by logistic regression. Followup eGFR data over 1.4 years from n=700 were available for analysis. AQP5 was undetectable in 138 patients. Tertiles of AQP5 were 0.4 [0 -2.2], 7.3 [5.9 -9.1], and 16.0 [13.0 -21.6] (ng/mL), respectively (p<0.01). Patients in the highest tertile of AQP5 had significantly higher total cholesterol, lower baseline eGFR, and higher levels of albuminuria compared to the lowest tertile. AQP5 was inversely correlated with eGFR slope (Pearson r = -0.12, p<0.001), and independent of clinical risk factors age, sex, race, and baseline SBP, DBP, HbA1c, Tot. Cholesterol, eGFR, and UACR (β = -0.05, p<0.004). Furthermore, AQP5 was significantly associated with fast eGFR decline (OR = 1.03 (95%CI 1.003 -1.06), p<0.03). Our data suggest that baseline AQP5 is independently associated with the progression of eGFR decline in patients with type 2 diabetes and nephropathy.Running title: Urine AQP5 as a prognostic marker of diabetic nephropathy.