RESUMEN
OBJECTIVE: To assess how mothers classify foods and how their eating practices interact with these classifications, with special attention to meanings and uses given to ultra-processed foods. DESIGN: Qualitative research with in-depth interviews and pile sorts. SETTING: Urban Brazilian Amazon. PARTICIPANTS: A sample of 34 mothers were selected through theoretical sampling. ANALYSIS: Content analysis for in-depth interviews and multidimensional scaling and cluster analysis for pile sorts. PHENOMENON OF INTEREST: Food classification. RESULTS: Classifications were based on context (ie, a time or a situation in which the food is eaten) and foods' healthiness. Five food groupings based on mothers' classifications were defined: (1) main meal foods, (2) fruits and fruit juices, (3) convenient foods, (4) leisure foods, and (5) canned sardines. Ultra-processed foods were classified differently from non-ultra-processed foods and considered unhealthy, consumed on special occasions or when there was no time or desire to cook. CONCLUSIONS AND IMPLICATIONS: Results highlight the potential of incorporating context-based categories and personal experiences to guide nutrition interventions and the potential of pile sorts to tailor messages to target populations.