Lung biopsies showing granulomatous disease are commonly sent for expert pathology consultation. On the basis of features we and others have identified, an algorithmic approach to diagnosis of these cases was developed. We hypothesized that applying this approach would increase the likelihood of rendering a more specific diagnosis, or by rendering either a narrower or broader differential diagnosis, offer a more clinically useful diagnosis. One hundred consecutive lung biopsies from patients with granulomatous and giant cell reactions were retrieved from our consultation files. Cases were categorized into those in which a confident diagnosis was made at sign out, ones in which a specific diagnosis was strongly favored, and those in which a differential diagnosis was suggested. One year later follow-up information was obtained and consultation diagnoses were compared with clinical diagnoses to determine the reliability of the approach. A confident diagnosis was rendered in 27 cases, a specific diagnosis was strongly favored in 34, and in 39 a differential diagnosis was provided. Consultant diagnoses were more specific in 47 of 75 (63 %) cases. In 15 cases, the differential diagnosis was expanded. The most common unrecognized diagnosis was aspiration pneumonia and the most common diagnosis omitted from the differential diagnosis by the primary pathologist was hypersensitivity pneumonia. Follow-up in 49% of cases in which it was sought, confirmed the consultant's diagnosis or was inconclusive in 97% of cases. The use of a standardized algorithmic approach to the interpretation of granulomatous disease in lung biopsies yields more specific and clinically useful diagnoses for consideration.