Questionnaires to assess student satisfaction with the work of university teachers have been used as tools to evaluate the quality of their performance and as evidence of teaching quality in institutional accreditation processes. The aim of the study is to compare the psychometric analysis of a Likert-type questionnaire on university teaching activity, using the Classical Test Theory (CTT) and Samejima's Graded Response Model (GRM). A database was available, containing the responses of students from different university courses belonging to a Spanish university. The analysis was performed initially using the TCT and subsequently with the MRG model. The parameters of the items were estimated and analyzed with both models and the information obtained with each one was highlighted. In conclusion, this application allowed identifying that the questionnaire has adequate psychometric properties that allow, with a high level of reliability and discrimination, to assess student satisfaction with teachers. Finally, the benefits of using the MRG for this type of instrument are presented, as well as the additional and detailed information provided by its use in the context of educational evaluation.