Introduction: As part of IPD data analysis of racial/ethnic disparities in Heart Failure we sought to harmonize sociodemographic, lifestyle, physiologic, biomarkers, and co-morbid covariates and HF outcomes. This report shares our data harmonizing strategies, matching characteristics, common formats and compares for diabetes, the cohort specific (unmatched) prevalence and prospective association with heart failure and compares them to the harmonized variable (common format). Methods: Nine cohorts (ARIC, JHS, WHI, Health ABC, MESA, FOHS, CARDIA, REGARDS, CHS) data is being harmonized for 128 covariates and HF outcomes of HF, HFrEF, HFmrEF, HFpEF. Twelve baseline variables ( age, sex, cigarette smoking, systolic blood pressure, diastolic blood pressure, pulse, height, weight, BMI, waist circumference, glucose, ECG, used identical formats and were completely matched. Seventy nine baseline variables (race, ethnicity, education, income, sleep, alcohol, CHD, DM, PAD, atrial fibrillation, etc) partially matched and a common format allowed for harmonization. Seventeen covariates (diet and physical activity) that measured the same construct but the questionnaires used differed enough that equipercentile equating was used to harmonize data. Results: In 64,114 participants from 5 cohorts, age range 20-95, mean age 62.7, the prevalence of diabetes mellitus was 10.5% using the common format, and 13.5% using cohort specific definitions. Prevalence varied by cohort from 7.9% to 23.4%. The harmonized prevalence differed by race and sex. The age, sex and race adjusted odds ratio for the prospective association with heart failure was OR=2.8, 95% CI 2.6,3.0. Sex and race specfic odds ratios differed by sex with females having a higher odds ratio, but were similar by race. Conclusion: Risk factors for heart failure are harmonizable in NIH CVD cohorts. Most variables are partially matched and require re-coding using a common format. Prevalence of DM varies by race and sex but association with HF only differs by sex.