This study estimated the levels and differentials in nutritional status and dietary intake and relevant knowledge of adolescent girls in rural Bangladesh using data from the Baseline Survey 2004 of the National Nutrition Programme. A stratified two-stage random cluster-sampling was used for selecting 4,993 unmarried adolescent girls aged 13-18 years in 708 rural clusters. Female interviewers visited girls at home to record their education, occupation, dietary knowledge, seven-day food-frequency, intake of iron and folic acid, morbidity, weight, and height. They inquired mothers about age of their daughters and possessions of durable assets to divide households into asset quintiles. Results revealed that 26% of the girls were thin, with body mass index (BMI)-for-age <15th percentile), 0.3% obese (BMI-for-age >95th percentile), and 32% stunted (height-for-age < or = 2SD). Risks of being thin and stunted were higher if girls had general morbidity in the last fortnight and foul-smelling vaginal discharge than their peers. Consumptions of non-staple good-quality food items in the last week were less frequent and correlated well positively with the household asset quintile. Girls of the highest asset quintile ate fish/meat 2.1 (55%) days more and egg/milk two (91%) days more than the girls in the lowest asset quintile. The overall dietary knowledge was low. More than half could not name the main food sources of energy and protein, and 36% were not aware of the importance of taking extra nutrients during adolescence for growth spurt. The use of iron supplement was 21% in nutrition-intervention areas compared to 8% in non-intervention areas. Factors associated with the increased use of iron supplements were related to awareness of the girls about extra nutrients and their access to mass media and education. Community-based adolescent-friendly health and nutrition education and services and economic development may improve the overall health and nutritional knowledge and status of adolescents.