Purpose <p>Nutritional epidemiological research is shifting its focus from individual nutrients to dietary patterns, which challenges traditional statistical methods. Here, we aim to apply various machine learning algorithms to identify and predict dietary patterns in the Dutch population.</p> Methods <p>Data on food consumption, sociodemographic and lifestyle factors from 867 males and 866 females participating in the Dutch National Food Consumption Survey (DNFCS) were analysed. K-means, K-medoids, and hierarchical clustering were compared to identify dietary patterns by sex. Six classifiers (naïve Bayes, K-nearest neighbours, decision tree, random forest, support vector machine and xgboost) were used to predict identified dietary patterns based on sociodemographic and lifestyle factors.</p> Results <p>After comparison, the optimal clustering method, K-means clustering, identified two distinct dietary patterns for both sexes, i.e. Traditional and Health-conscious patterns. The Traditional pattern was characterised by a higher energy intake and consumption of bread, potatoes, red and processed meat, coffee, fats and oils, and sugary drinks. Conversely, a higher intake of fruit, vegetables, tea, nuts, seeds, and breakfast cereals characterised the Health-conscious pattern. The classification models demonstrated moderate predictive accuracies (60–68%). According to the classifiers, the most important predictors for both sexes were education level, age, and BMI.</p> Conclusion <p>Machine learning algorithms can be useful in identifying dietary patterns in population studies. We identified Health-conscious and Traditional patterns in a Dutch population, suggesting tailored public health interventions towards individuals adhering to a Traditional pattern. Future research should improve model validity and reproducibility to enhance its applicability in public health interventions and dietary guidelines.</p>