Objectives: Eating difficulties are common for patients in hospitals (82% have one or more). Eating difficulties predict undernourishment, need for assistance when eating, length of hospital stay and level of care after hospital stay. Eating difficulties have through factor analysis (FA) been found to belong to three dimensions (ingestion, deglutition and energy). The present study investigates inter-observer reliability. Other questions at issue are if the findings from the previous FA can be confirmed, if adjustments need to be done and if the Minimal Eating Form (MEOF) can serve as an assessment model for identification of eating difficulties. Previously found associations between eating difficulties and outcomes as well as measures taken to improve oral intake were also investigated. Design: Inter-observer study and cross-sectional observational study. Settings: Hospitals and special accommodations (SAs). Participants and measurements: Inter-observer study: Observers made standardized assessments of eating, independently and at the same time, on a sample of 50 patients with stroke. Survey study: 2600 (88%) out of 2945 persons agreed to participate in a survey of eating and nutrition. All SAs within six municipalities and six hospitals were involved. Nursing students, clinical tutors and staff performed the assessments, supported by the researchers. Results: The average agreement between observers of eating difficulties was 89% (Kappa coefficient 0.70). In the survey study, the mean age of persons (n=1726) living in SAs was 85 years (SD 8) and 69% were women, while the corresponding figures for patients (n=874) in hospitals were 69 years (SD 18) and 53% women. Low Body Mass Index (BMI) was found in 27%, unintentional weight loss in 23% and need of eating assistance in 38% of the persons. Protein- and energy- (PE-) enriched food was given to 4%, adapted consistency of food to 23% and food supplements to 16% of the persons. The new FA confirmed the previous one and minor adjustments of the model were made. Having ingestion difficulties was the strongest predictor of need for eating assistance (OR 14.5). Deglutition difficulties strongly predicted serving of adapted consistency of food (OR 7.3). Poor energy levels and reduced appetite predicted weight loss (OR 6.0), BMI below limits (OR 2.5), supplements (OR 5.3) and PE-enriched food (OR 3.4). Conclusions: The MEOF has satisfying validity and reliability. The earlier model of eating difficulties was confirmed (MEOF-I), and the model was slightly adjusted to a new model, MEOF-II. Providing eating assistance seems effective in preventing malnutrition (weight loss and BMI below limits), and is mainly provided to persons with ingestion difficulties. Difficulties with energy intake and appetite are not associated with eating assistance; indicating that those persons might need support of some other kind. This support can include providing PE-enriched food and supplements, but seems however insufficiently or inadequately delivered, as low energy and appetite problems are also associated with both weight loss and low BMI. Findings from other studies are confirmed.