Complexity is a very broad subject that applies to project management, engineering design and manufacturing, arithmetic, software, statistics, etc. In maintenance systems, complexity can be defined based on technical and managerial aspects of a maintenance project. Because relative complexity between two projects can be used as a yardstick for resource allocation between them, quantifying the complexity becomes important. To quantify the complexity of maintenance projects, this thesis reports two models.In uncertain situations, a fuzzy graph-based model is developed that determines relative complexities of maintenance projects based on experts‟ opinions with respect to technical and managerial aspects. These aspects may not be measured precisely due to uncertain situations. The model uses an aggregation operator to mitigate conflict of experts‟ opinions on complexity relations. Using a fuzzy relation matrix representing the degrees of membership of relative complexities, the model maps the fuzzy graph into a scaled Cartesian diagram.Also, complexity of a maintenance project can be investigated through time to repair (TTR). Performing statistical analysis shows that human cognition and project complexity have significant influence on TTR. These influential factors can be studied by a learning curve. Due to the nature of maintenance calls for repairs, a learning curve model made up of two segments is proposed. A project complexity can be derived from the learning curve at the breakpoint time. Taking into account human cognitive abilities, the breakpoint indicates the required number of trials in order to reach mastery level for performing certain tasks unsupervised.