In this thesis, a thermal comfort control for trams is developed, which minimises the power consumption of the heating, air conditioning and ventilation system. The objective of this thesis is to achieve a satisfying indoor climate in a tram with the smallest electrical effort possible, by using methods of control engineering and closely related sciences. In this work, mathematical models based on the fundamental laws and balance laws of thermodynamics of all thermally relevant parts are stated. Individual parts modelled are, the interior of the tram, the heating, ventilation and air conditioning system (HVAC), and the controller to calculate the temperature, relative humidity, and carbon dioxide concentration in the interior of the tram. For the model of the interior, a macroscopic modelling approach is chosen. For the HVAC two different approaches will be presented: The first approach is an analytical modelling of the subsystems (mixing chamber, cooling cycle and ventilators etc.) to depict the entire system. The model of the entire system is non-linear. For the second approach, various simplifications are used during the modelling to gain a linearised model. The controller will be set up as a finite state machine. In the next step, the numerical values of the parameters for the created models will be estimated. For the interior model of the tram, parameters are determined by a least squares approach from measurement data. Necessary data are gained from experiments carried out in the climatic wind tunnel (CWT). If expert knowledge about the parameters is available, it will be used as a basis for further adaptation. In the Heating, Ventilation and Air-Condition Unit, the parameters will be derived from design sketches and similar documents. If there is no or too little information, parameters will be estimated by grey-box approaches. For the controller model, the states and transition conditions of the finite state machine will be made available by the manufacturer. The estimated parameters will be validated by data, which will be collected during CWT and on-site experiments. From literature, a common model of thermal comfort is taken and numerically linearised. With the linearised models of the plant (interior of the tram and HVAC unit) cascaded control loops are established. The master control loop, whose controller is a model-predictive controller, regulates the thermal comfort of the tram. The controller of the slave control loop will be realised with a mixed-integer optimisation, which converts the set point of the auxiliary control variable in an energy-optimal way. For the mixed-integer problem, a heuristic solution is used. For confirming the solution approach, the algorithms were implemented on a rapid controller prototyping platform. First, the control concept was tested during several CWT tests. The annual energy consumption was extrapolated using CWT measurements.