To improve understanding of the clinical characteristics and diagnosis of hypersensitivity pneumonitis (HP).We retrospectively analyzed the clinical data, including clinical symptoms, laboratory tests, exposure, pulmonary function tests, chest CT imaging and cytological classification of bronchoalveolar lavage (BAL) of 96 patients with HP from Jan 2001 to Jun 2011 in Peking Union Medical College Hospital. We divided the patients into 2 groups: a pathologically-confirmed group and a clinically-suspected group.There were 58 females and 41 males. The median age at the diagnosis was 53 years. The most common exposures were low-molecular-weight chemicals (42.7%) and animal proteins (37.5%). Common clinical symptoms included dyspnea on exertion (90.6%) and cough (76.0%). Pulmonary function test showed diffusion abnormality (73.5%) and restrictive ventilatory impairment (59.7%). Chest CT scan revealed patchy or diffuse bilateral ground-glass opacities (64.6%), centrilobular nodules (21.9%), and air trapping (15.6%). Reticulation (45.8%), traction bronchiectasis (21.9%) and honeycombing(9.4%) were present in chronic HP. BAL lymphocyte counts0.2 and CD4/CD80.9 were more commonly seen in patients with a disease course of less than 1 year. The pathologically-confirmed group and the clinically-suspected group shared many similar characteristics including age at diagnosis, gender, clinical manifestation, pulmonary function impairments and imaging findings, but significant differences existed in certain parameters. In the pathologically- confirmed group, the duration of disease was longer (24 months vs 6 months, Z = -2.492, P = 0.013) and clubbed fingers were more common (23.4% vs 8.2%, χ(2) = 4.227, P = 0.040). Diffusion abnormality was present in more patients of this group (90.7% vs 44.0%, χ(2) = 35.219, P0.01). By CT scan, reticulation, traction bronchiectasis and honeycombing (57.5% vs 26.5%, χ(2) = 9.434, P0.01) were more evident as compared to the clinically-suspected group. The value of transbronchial lung biopsy for diagnosing HP was limited, with a positive result of only 8.2%. Surgical lung biopsy was needed in uncertain cases.The diagnosis of HP was difficult. In some cases a clinical diagnosis can be made by combination of history of exposure, CT manifestations and cell classification of BAL. For atypical cases a multi-disciplinary approach including pathologists, radiologists and pulmonologists is needed.