In many studies, correlations between nutrition knowledge and dietary behaviour have failed to reach statistical significance, leading researchers to question the relevance of nutrition knowledge to food choice, and the value of nutrition education campaigns. This study aimed to investigate the relationship between knowledge and intake of fat, fruit and vegetables using a well-validated measure of nutrition knowledge. The study was a postal survey, using 1040 adult participants selected at random from General Practitioners' lists in England. Nutrition knowledge and food intake followed the expected demographic patterns. Knowledge was significantly associated with healthy eating, and the effect persisted after controlling for demographic variables. Logistic regression showed that respondents in the highest quintile for knowledge were almost 25 times more likely to meet current recommendations for fruit, vegetable and fat intake than those in the lowest quintile. Nutrition knowledge was shown to be a partial mediator of the socio-demographic variation in intake, especially for fruit and vegetables. This demonstrates the value of using more sophisticated statistical techniques to investigate associations between knowledge and food intake and indicates that knowledge is an important factor in explaining variations in food choice. The results support the likely value of including nutrition knowledge as a target for health education campaigns aimed at promoting healthy eating.