Aims: The aim was to test the expression of PU.1 on different types of histiocytoses and to test the utility of PU.1 in confirming or excluding a histiocytic origin in tumour samples with suspicion of histiocytosis. Methods and Results: We analysed 66 biopsies of nonmalignant histiocytoses represented by Langerhans-cell histiocytosis (n = 13), Erdheim–Chester disease (ECD) (n = 19), Rosai–Dorfman disease (RDD) (n = 14), mixed ECD-RDD (n = 3), ALK-positive histiocytosis (n = 6), and juvenile xanthogranuloma (n = 11). All cases were positive for PU.1 in reactive and neoplastic histiocytes. In addition, 39 cases of tumours with high-grade cytological atypia were referred to our center as suspicion of malignant histiocytosis/histiocytic sarcoma and only 18 were confirmed. Indeed, more than half of these tumours (21/39) were either undifferentiated malignant tumours with a stroma rich in histiocytes, diffuse large B-cell lymphoma, or high-grade dedifferentiated liposarcoma. PU.1 was useful to distinguish between the negativity of large atypical nuclei and the positivity of stromal reactive histiocytes. Conclusion: PU.1 is expressed by all types of histiocytosis. It distinguishes histiocytosis from histiocyte-rich tumours with an easy interpretation due to its sharp nuclear staining. Its negativity in lesional/tumour cells in histiocyte-like lesions is useful to eliminate a histiocytosis.