Introduction: The primary objective of this study is to determine the body composition of Colombian women and its relationship with mammographic density. Methods: Cross-sectional descriptive study with an analytical component. Using non-probabilistic sampling, 670 patients who met the inclusion and exclusion criteria were selected. Qualitative variables are represented by absolute frequency distribution and proportions, and quantitative variables by central tendency and dispersion statistics. A simple linear regression model was fitted to determine the relationship between self-reported and measured values of weight and height, as well as the relationship between indicators of adiposity with age, contraceptive use, and menopausal status. Also using a simple and multiple linear regression model, mammographic density is related to clinical and anthropometric variables, taking a significant value of p<0.05. Finally, a decision tree was built to identify those explanatory variables to be included in a final regression model to be able to estimate mammographic density, to which the statistical power of the model was finally calculated to assess its possible utility in clinical practice. Results: The mean age was 57.9 years (SD 6.3). According to the BMI, 43.4% were overweight, followed by normal weight, obesity grade I, grade II, grade III and underweight, with percentages of 29.1%, 16.6%, 6.3%, 1.2 % and 0.3%, respectively. Older women had a higher percentage of fat mass (p 0.02) and waist/hip ratio (p 0.01). 53.4% of the women had a mammographic density of less than 25%. Premenopausal women presented 14.2% higher mammographic density than postmenopausal women. BMI and age presented an inverse linear relationship with mammographic density (B= -4.0, p value 0.000; B= -0.7, p value 0.000, respectively). Age maintains an inverse linear relationship with the percentage of mammographic density, while oral contraceptives showed a direct linear relationship, both statistically significant. Age, BMI, waist circumference, and bra size were identified as variables likely to be included in a mammographic density percentage prediction model. Conclusions: The highest proportion of women in this study were in the overweight range, as well as with low mammographic density patterns (<50%). It was possible to propose the inclusion of variables such as age, BMI, bra size, and waist circumference in a final mammographic density estimation model, estimating a statistical power of 1.00 for the model.
