This study aimed to compare different methods of assessing dietary exposure to flavourings in the context of a stepwise approach. The dietary exposure to four flavourings--raspberry ketone, glycyrrhizinic acid, coumarin, and caffeine--was determined. When dietary exposure exceeded the safety limits, the need for more detailed assessment using less aggregated data was judged necessary. First, screening methods--maximized survey-derived daily intake (MSDI), single-portion exposure technique (SPET), and modified theoretical added maximum daily intake (mTAMDI)--were applied. Next, individual food consumption data were used for creating models with different levels of detail to identify the foods: a model based on food groups and models based on food items. These were collected from 121 Dutch adults using a standardized 2 x 24-h dietary recall (EPIC-Soft) in the European Food Consumption Validation (EFCOVAL) study. Three food item models were developed: without improvements of the flavouring descriptor built in the software; with improvements; and with use of non-specified flavour descriptors. Based on the results of at least one of the three screening methods, refined assessment was necessary for raspberry ketone, glycyrrhizinic acid, and caffeine. When applying the food group model, the need for refinement was indicated for the four flavourings. When applying the food item models, only glycyrrhizinic acid and caffeine presented dietary exposure above the safety limits. In the raspberry ketone case, dietary exposure increased when improvements in food description were considered. The use of non-specified flavour descriptors hardly changed the results. The collection of detailed food consumption data at the individual level is useful in the dietary exposure assessment of these flavourings.