The process of designing cementitious layers (weakly and strongly cemented) against fatigue distress in road structures is well accepted. Research and field investigations with the aid of the Heavy Vehicle Simulator (HVS) revealed, however, that almost all weakly cemented subbase layers undergo non-traffic and traffic¬associated cracking and eventually degradation of the cemented material into a granular state (post-cracked phase). It is therefore very important to analyse these layers in the post-cracked phase and to incorporate the results of this analysis in the design, for both new and rehabilitation designs. The investigations revealed that the rate of degradation of these materials is largely dependent on traffic loading and the moisture conditions within the pavement layers. The purpose of this study is to investigate the behaviour of weakly cemented subbase layers in road structures mainly under a bitumen base between 90 mm and 140 mm thick. This behaviour includes both pre-cracked and post-cracked phases. It is shown that the fatigue life of bitumen base layers is mainly governed by the condition of the weakly cemented subbase layers. In Chapter 1 a brief historical review is given of the development of fatigue distress criteria of the cementitious layers. It is shown that the maximum horizontal tensile strain at the bottom of these layers is the main distress criterion in the pre-cracked phase. Unconfined compressive strength and durability requirements are also discussed. Some aspects of the current design methods are outlined in Chapter 2. The concept of equivalent granular states in the post-¬cracked phase of cementitious layers was derived from HVS test findings. However, before this document no behavioural prediction models were available to quantify accurately the post-cracked state of these layers. The actual mechanisms of distress were also not clear. In Chapter 3, a detailed investigations and analysis of ten dif¬ferent HVS tests at four different sites in Natal are discussed. The purpose of the analysis, is firstly to illustrate the powerful method of full-scale accelerated HVS-type testing and secondly to indicate the importance of the upper subbase layer, the initial condition of the in-situ structure, the importance of water condi¬tions within the pavement structure, and finally the different states of behaviour of this type of road structure, including predictions of future behaviour based on linear elastic theory. The characteristics of the weakly cemented upper subbase layer are shown to be of paramount importance in the final behaviour of these structures. In Chapter 4 a method of analysing the behaviour of mainly weakly cemented layers in the post-cracked phase is proposed. This method arises from the HVS testing discussed in Chapter 3, and may be regarded as the most important improvement on the current method discussed in Chapter 2. The analysis incorporates the determination of the effective elastic moduli of weakly cemented subbase layers, including both the wet and the dry periods during the structural design period of these layers. In Chapter 5 the effect of relatively weak interlayers within asphalt base structures is discussed and evaluated. The analysis incorporates the relative position and thickness of the inter layer during both wet (low modulus) and dry (high modulus) conditions. A summary and detailed discussion, together with recommendations for future research, are given in Chapter 6. The need for the incorpo¬ration of durability (erodibility) criteria for weakly cemented materials is also discussed. More research should be done on the effects of accelerated curing compared with normal curing methods. This investigation includes aspects of soil-lime-cement reactions together with delayed compaction techniques to reduce shrinkage cracking. The need for better quality control as well as improved construction techniques for weakly cemented materials is also discussed. This thesis also contains two appendices. In the first of these detailed photographic records of the different HVS tests and performances are given. In the second appendix an example of an input computer program to plot the three dimensional behavioural model is given.