An observation protocol was created and used in order to observe the instruction process of the 14 teachers who participated in the study, so as to ensure the consistent implementation of differentiated instruction by all teachers and in order to identify the characteristics of their instruction. The observation protocol was created based on the main characteristics of differentiated instruction according to the literature review and consisted of 18 criteria on a 5-point Likert scale. For the validation of the observation protocol, initial values for each criterion were equated through a Rasch model and specifically the ‘‘Rating Scale’’ model (Andrich, 1978), in order to equate the degree of difficulty for the implementation of the criteria between the first and second observation, thus determining the degree of difficulty. The values of the model, mean infit, mean square and mean outfit were very close to one (1) and values for infit and outfit ranged from - 0.08 to 0.08, thus giving validity and reliability to the observation protocol. Based on these data, the observation protocol comprises a valid and reliable research tool that can be used for the observation of differentiated instruction. Simultaneously, these results answer one of the study’s aims by describing the ways and the degree to which differentiated instruction is implemented by teachers in order to meet the diverse needs of their students.