This piece of research is an attempt to demonstrate that designing at a regional scale can bring legibility and contrast to the landscape. This in contrast to the current mode of urbanization in which designing is done at housing development scale thereby banning from the landscape the larger dimension. This theory we examined by way of three models. Each model proceeds from extremes in the degree of scatter and concentration as well as from built and unbuilt surface area. The programme is fixed in all three. 1. The concentration model combines a compact city of 230,000 inhabitants in the south of Hoekse Waard, a major island south of Rotterdam, with a large-scale agricultural area the size of the island of Texel. 11. The scattered concentration model consists of eleven elongated villages one kilometre wide and each with a population of 20,000. A necklace of parks and natural lakes is slung between the villages. To reach a park you need walk no more than 500 metres from your house. The house you live in has a garden and a garage. 111. In the scatter model the landscape is your garden. You live in ribbons with a combined length of 500 kilometres. The 'garden' is managed by eco-farmers. Businesses, glass nursery houses and shopping malls lie in supercomplexes along the railways and motorways. The three models have been tested against five criteria: exploitation of the potential of the infrastructure; deployment of the large dimensions and scale of Hoeksche Waard; whether organized to take advantage of combinations of programmes; whether able to accept further growth as well as stagnation; whether there is regional orientation. The conclusion is that 1 and 111 offer the widest perspective for Hoeksche Waard. By placing the city of model 1 in the waters of Haringvliet southwest of the island, the city acquires a high-powered identity. Concentrating programmes engenders critical mass which in turn produces new programmes and specific solutions and guarantees a vibrant urbanity. Model 11 generates little in the way of new programmes through its lack of a broad basis. In model 111, the developments are contingent upon each other, making this model extremely vulnerable where phasing is concerned. All three models show clearly that deploying the regional scale can produce legible schemes out of which emerge new combinations of built and unbuilt parts. Institution: Rotterdam Tutors: Harm Veenenbosch, Rients Dijkstra, Dirk Sijmons, Ron van Genderen & Harm Tilman Specialization: urban design