The tropospheric delay is one of the main error sources that affects the quality of GNSS positioning. Since modelling is the only way to mitigate the tropospheric delay, a proper model needs to be applied during GNSS data processing. Several different tropospheric delay models are usually implemented in modern GNSS post-processing software, so users will often not be sure which one to use. The answer to this question represents the main goal of this master's thesis. Six different tropospheric delay models that are implemented in Leica Infinity software were tested with respect to obtained GNSS baselines quality. Our analyses are based on several different data sets, each containing thirty baselines. In order to find an optimal processing strategy to obtain the best achievable position quality, different tropospheric delay models were assessed on baselines acquired in different tropospheric conditions, different daytimes and with different time span of observations, different heights of baseline endpoints and different baseline lengths.