This is the first large-scale research in the Netherlands into the relation between neighbourhood characteristics and exercise behaviour of residents. The study maps the exercise behaviour of residents of four Amsterdam neighbourhoods. There are clear differences in urban planning in these neighbourhoods, whereas residents all have comparable low socio-economic status. First of all the behaviour of 443 adults and 193 children in four Amsterdam neighbourhoods has been recorded by means of questionnaires. These neighbourhoods are comparable in population composition (average income, ethnic origins etc.) but differ in the degree of built area (such as amount of greenery and urban density). Residents were asked how they reach destinations inside and outside their neighbourhood: how they run errands, go to work and by what means they go to the park. They were also asked how they value a number of neighbourhood characteristics, like social and traffic safety, greenery in the neighbourhood and the appearance of their environment. In the second place the neighbourhoods have been spatially analysed on the basis of the following characteristics: urban density, presence and location of services, particularly shops for everyday purchases and parks. In addition the quality of the routes to these services has been investigated, as well as the accessibility of the neighbourhoods by public transport and parking. The results of the study demonstrate a clear difference in everyday exercise patterns by the residents of a classic urban district like Boerhaavebuurt and post-war neighbourhoods in Osdorp. Residents of the classic urban district prove to walk and cycle more on a day-to-day basis than residents of the post-war neighbourhoods. The garden village-like Pekbuurt takes up an interesting intermediate position. The difference in exercise behaviour between the neighbourhoods is also apparent in the differences in (average) BMI values (Body Mass Index, a measure for the degree of obesity). In neighbourhoods where people exercise more, the number of people with (morbid) obesity is lower. We assume that the differences in exercise behaviour and the related BMI values are in part due to the planological construction of the neighbourhoods.