Lake Van is located in the eastern part of Türkiye and forms the largest soda lake in the world. In this study, we present the relationship between instrumental data which belongs to Lake Van level changes and meteorological parameters by performing a multilevel analysis. The data set consists of monthly average levels of Lake Van and monthly meteorological parameters (temperature, precipitation and wind speed) between 1944-2019 years. Two different multilevel linear models; random intercept, random intercept and slope model were used. Unstructured (UN) covariance structure and Maximum Likelihood (ML) were used for repeated measurements and estimation, respectively. Log-likelihood (ll), Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC) and AIC Corrected (AICC) were used for the selection of the best model. The random intercept and slope model (Model II) is explained as the best model for the lake level changes in this study. Statistical results in this study indicated that the temperature and wind speed are the key parameters controlling the Lake Van water level fluctuations, whereas the precipitation effect is minimal due to the type of precipitation (snowfall). For this reason, temperature, wind speed and also the type of precipitation (snowfall or rain) must be considered for disaster modelling in settlements of Lake Van and similarly closed basin lakes.