Objective: Description and application of an adapted semiquantitative food frequency questionnaire (SFFQ) for dietary assessment in the elderly population of the Rotterdam Study. Design: Dietary assessment consisting of a two-step approach was performed in 5434 participants (2225 men, 3029 women) of the Rotterdam Study from 1990 to 1993, a population-based prospective cohort of 7983 subjects aged 55-95 years (participation rate 78%). Statistical analysis: Nutrient intake was calculated for men and women in four age groups (55-64 years, 65-74 years, 75-84 years, 85-95 years) and linear trend analysis for differences in mean nutrient intake across age groups (55-64 gears 65-74 years, 75-95 years) by regression analysis was conducted. The influence of baseline characteristics on energy and nutrient intakes adjusted by age and sex was investigated by one-way-analysis of variance. Results: The adapted SFFQ made it possible to measure nutrient intake in the elderly within a limited time frame (2 x 20 min) across a wide age range (55-95 years). For nutrient intake we observed a general decline in mean intake of energy and most nutrients with age in men. In women the relation with age was not consistent: for most nutrients mean intake showed a decrease with age (e.g. water, magnesium, potassium), for some an increase (e.g. total fat, saturated fat, mono/disaccharides), and some nutrients showed no substantial change (e.g. calcium, retinol). Reported nutrient intake was influenced by body mass index, smoking status, socioeconomic status and activities of daily living. A prescribed diet was reported by 12.9% of participants and 34.6% used supplements on a regular basis. Conclusions: The described two-step approach for dietary assessment in the elderly facilitated collection of data on dietary habits across a wide age range and within a limited time frame making it a suitable instrument for application in large-scale epidemiological studies in the elderly.