This paper deals with the fit indices used in Structural Equation Modelling (SEM) for testing theoretical models and the difficulties that can occur during the testing of theoretical models in different fields of psychology. The paper discusses the basic assumptions of SEM and presents the indices used for assessing the fit of theoretical models. This paper also presents the procedures for calculating the basic statistic for assessing the fit of models (χ2), as well as for calculating the most commonly used fit indices, in order to gain a better insight into the advantages and potential difficulties that can occur during their usage. We mention the difficulties regarding the assessment of fit of the model based on χ2 and the discussed fit indices stemming from the sample size, data distribution and assessment methods, wrong specification of model and disturbance of normality and independence of latent variables, as well as the ways in which these difficulties can be overcome. This paper provides a proposal for the approach to presenting the fit indices in reports on studies where theoretical models were tested via SEM.