There are mainly two concepts that appear with the development process of urban settlements. One of them is change and the other is sustainability or continuity. Unchanged and changed properties are strictly related to the development process itself and they differ in planned and self-generated systems. The present paper is intended to contribute to understanding the principles, which lead to the spatial evolution of cities through time. Beyazit square covering a 1km × 1km area, with its surroundings was selected for this study as it is one of the historical cores of Istanbul and has a unique character with its built environment and rich urban activities. Analyzing the changes in spatial structure in time shows the effects of the natural environment and the formerly built up environment as an initial condition. Although the general structure, which is shaped mainly by the dominance of historical environment, is unchanged, in the adaptation process for the changing social, economic and several other factors, many changes have occurred. This complicated process is examined by using the methods defined by chaos theory such as analyzing the changes in fractal dimensions in different time periods to explain the changing complexity level of spatial structure. The two key features of fractal dimension concept are generator and initiator. These features can be translated into urban pattern concept as the generation process or transformation rules and existing environment as an initial condition. Planning and selforganization represent two different processes leading to the transformation of the urban pattern. First of them is a topdown process, which has an ability to make changes on the whole urban pattern in a short time period, while second one reflects a bottomup process, where changes are made by individual decisions and they take longer to change the spatial pattern. Moreover, these two processes generally occur in different scales: Selforganization is closely related to buildings and plots while planning is generally more comprehensive and affects building blocks, transportation routes and the whole urban pattern. Because of these, the urban pattern can be classified into three main components: buildings, building blocks and roads. Fractal dimension in different time periods for each element is calculated separately. Although urban pattern is a fractal object, there is a big difference between computer generated theoretical fractals and urban pattern: Generally theoretical fractals are purely selfsimilar which is not applicable for real urban patterns. Added to this, urban pattern is a highly complex structure, the fractal dimension changes in different scales and different sub regions. Fractal generation in computer environment starts with the simple initiator object and generation rule, however, in urban realm, generally, the existing patterns are very heterogeneous and there are many interacted factors that affect the evolution process as a generation rule which result in various fractal dimensions. This feature is called as "multifractality". There are several methods to calculate fractal dimension such as: Hausdorff dimension, Self similarity dimension, Box counting dimension, Divider or ruler dimension, Lyapunov dimension, Information dimension, Dilation dimension. Box counting method is the most preferred to other methods in physical science. Box counting method is relatively easy to calculate and suitable to measure spatial elements of urban pattern. Because of these advantages, box counting method was selected to calculate the fractal dimension of uban pattern. The preliminary measurement results show that the fractal dimension of the Beyazit sample is over 1.7 that is higher than the values of 1.41.6 measured in modern settlement patterns. These results are considered to be a reflection of the complexity of the spatial structure of the study area. Detailed measurements of urban pattern revealed the multifractality of Beyazit sample. Beyazit has different fractal dimensions in different scales. The more diversity of spatial elements and heterogeneous structure of urban pattern bring together the increase in the range of fractal dimension values. On the other hand, fractal dimension values are generally over 1,7 and very close to each other with the exception of the year of 1819, which demonstrates very different pattern. Similar fractal dimensions of different time periods reflect the continuity of spatial character especially after 1900s. On the other hand, the changes in fractal dimension values are the result of spatial changes such as widened roads or changing building blocks and buildings demonstrating the relationship between fractal dimension and spatial configuration. [ABSTRACT FROM AUTHOR]