Background: The WATCH-DM and TRS-HFDM risk scores were developed to predict 5-year risk of incident heart failure (HF) among individuals with type 2 diabetes mellitus (T2DM). We externally validated both scores across different cohorts and clinical settings with varying baseline risk. Methods: Incident HF risk was estimated by the integer-based WATCH-DM(i) and TRS-HFDM scores in participants with T2DM free of baseline HF from two randomized clinical trials (TECOS, N=12,028 and Look AHEAD, N=4,867). The WATCH-DM(i) score was also validated in electronic health record (EHR) data from a single large healthcare system (N=7,475). Model discrimination was assessed by Harrell’s concordance index (C-index) and calibration by the Greenwood-Nam-D’Agostino (GND) statistic. Results: HF incidence rate was 7.5, 3.9, and 4.1 per 1000 person-years in the TECOS, Look AHEAD, and EHR cohorts, respectively. WATCH-DM(i) and TRS-HFDM scores had similar discrimination and calibration for predicting 5-year HF risk in the Look AHEAD cohort (C-indices = 0.70; GND p-value >0.30 for both). Both scores had lower discrimination and underpredicted HF risk in the TECOS cohort (C-indices = 0.65 and 0.62, respectively; GND p-value <0.001 for both). In the EHR cohort, the WATCH-DM(i) score demonstrated superior discrimination (C-index = 0.73) and no evidence of miscalibration (GND p-value = 0.96). Conclusions: The WATCH-DM and TRS-HFDM risk scores can discriminate risk of HF among intermediate-risk T2DM populations.