A three-year food behaviour study of non-institutionalized urban elderly aged 65 to 77 years, of British descent and living alone, included a nutrition intervention program designed to maintain and/or improve dietary practices. The effect of the program was evaluated by changes in knowledge, perceptions of 14 "marker" foods and use frequency of 194 foods. The design was one of the pre- and post-test comparisons between an experimental and control group. Although the findings admit the difficulty in changing dietary practices of seniors, they do provide evidence that some changes are feasible at the cognitive level and in the perception of foods. They document continuous changes in seniors' nutrition behavior, which, though difficult to assess, requires the attention of nutritionists.