Objectives This study aims to determine the association between parosmia and clinically relevant recovery in olfactory function in patients with smell loss receiving olfactory training. Design and setting This was a retrospective cohort study of patients that received olfactory training. Adult patients with the major complaint of quantitative smell loss were recruited and treated at several ENT clinics in German between 2008 and 2018. Participants A total of 243 participants were included. Main outcome measures Changes in olfactory function after olfactory training. Age, gender, baseline olfactory function, etiology and duration of smell loss, duration of training, and presence of parosmia and phantosmia were assessed for their impact on clinically relevant changes in overall and subdimension olfactory function using binary logistic regression analysis. Results Relevant improvements in discrimination function were more likely in those that had lower baseline olfactory function, postinfectious reasons compared to posttraumatic or idiopathic causes and those that had parosmia at initial visit. Relevant improvements in odour identification were more likely in those that had a lower baseline olfactory function, female gender, and in those who had parosmia at the first visit. Clinically significant improvements in odour threshold were more likely in postinfectious causes compared to posttraumatic reasons and those who were older in age. Conclusions This study demonstrated that the presence of parosmia is associated with clinically relevant recovery in olfactory function in patients with smell loss receiving olfactory training.