Autism Spectrum Disorder (ASD) is a complex condition that falls under the category of so-called Neurodevelopmental Disorders. It is characterized by the presence of dysfunctions in social interactions, in the communicative capacity, symbolic thinking, as well as, stereotypical behaviours and deregulation of sensorial character. Currently, given its high prevalence, it has awakened scientific interest in order to carry out earlier diagnoses that will have an impact on more effective interventions.Until now, the diagnosis of ASD has been carried out by means of standardized diagnostic batteries based on qualitative criteria, so that the response may be far from that obtained in a real context.Therefore, tools such as Virtual Reality (VR), with the potential to reproduce highly realistic environments, can be a valid context for both assessment and therapeutic intervention. Non-immersive virtual environments have been applied, given the refusal of the ASD population to use VR viewers, although with observational results. These limitations can be overcome with the use of so-called Immersive Virtual Environments (IVEs), since they represent a non-invasive technological solution, with a greater capacity for immersion and, therefore, for generating responses that are more similar to those obtained in a real context.Observational measurements can be overcome with physiological measurements, such as electrodermal activity (EDA), which provides the body's response in the form of sweating to a stimulus, or eye tracking, which shows the behaviour of the eye. Both involve implicit, unconscious and quantifiable responses, which can help define the condition.Consequently, the present Doctoral Thesis, composed of three studies, aims to unite the use of IVE, with the capacity of visual, auditory and olfactory stimulation with physiological measures, focused on the evaluation and training of ASD, as well as studying the relationships between them and the diagnostic batteries of ASD.