Modelling of cancer is a long-standing problem. The Gompertz-law can be adapted to several tumour growth dynamics with high accuracy. However, physiological models are difficult to construct because of the complexity of the surrounding tissue and the tumour itself. The restricts us to small steps in tumour growth modelling, and to a reductionalist approach. Two models for avascular in vitro tumour growth are developed and compared for the suitability and their limits to describe the data. One model is agent-based and respects individuality and mechanics of cells, while the second model is a partial differential equation model. It turns out that both models are equally able to describe the tumour growth data, but that there are crucial differences in the morphology of the modelled tumours.