This study used a correlational and cross-sectional research design to identify factors influencing the nutritional status of a sample of elderly Taiwanese. Guiding the study was a conceptual framework that integrated self-efficacy theory and health locus of control theory. Bivariate correlations among the study variables were examined, and hierarchical multiple regression tests were performed to determine whether and to what degree background characteristics, nutrition self-efficacy, and health locus of control predicted nutritional status. Furthermore, the study analyzed and categorized topics identified in the participants' verbal explanations of their eating patterns. Of the 156 Taiwanese elderly who completed the survey, the majority were healthy males (60.9%) with a mean age of 72.29 years. Relationships among the variables indicated that older people had lower nutrition self-efficacy and nutritional status. Females tended to have higher chance health locus of control. Higher educational levels were related to better nutrition self-efficacy and Mini-Nutritional Assessment scores. Moreover, elderly persons who reported greater health problems and medications had lower Mini-Nutritional Assessment scores, but had higher albumin levels. The results also showed that, among the elderly, nutrition self-efficacy, internal and chance health locus of control, and nutritional status were interrelated, whereas the powerful-others orientation was not correlated with other variables. Nutritional status, age, educational level, health problems and medications, and chance health locus of control all had effects on the Mini-Nutritional Assessment scores, and only health problems and medications explained significant variances in albumin levels. In the qualitative portion, a subsample of 58 participants (37.18%) answered an open-ended question about their eating patterns. The analysis of their responses revealed three categories of eating patterns: eating and old age, eating and faith, and eating and family harmony, with each encompassing a number of subcategories. Generally, the study confirmed the relationships proposed in the research model; however, the overall amount of predicted variance accounted for by the predictors was small, which indicated that unmeasured factors might account for most of the variance in nutritional status. Further research is therefore necessary to gain a deeper understanding of nutritional status and its factors among the Taiwanese elderly.