Background and Design.—Rashes in immunocompromised cancer patients can be important, and skin biopsies are often recommended for their evaluation. The objectives of this study were to determine how often skin biopsy in these patients is performed and how often it alters diagnosis and therapy. Records of all immunocompromised adults with cancer and acute rash seen by dermatology consultants on a hematology-oncology ward of a university hospital for 39 months were reviewed to determine patients' course and outcome (190 episodes of rash in 123 patients). Results.—Skin biopsies were performed on 108 rashes (57%); 82 rashes (43%) were evaluated without biopsy. Among the 108 patients who underwent a biopsy of their rashes, the biopsy findings supported the prebiopsy diagnosis in 51% (95% confidence interval [CI], 42% to 60%), altered it in 44% (95% CI, 35% to 53%), and did not contribute to the final diagnosis in 6% (95% CI, 2% to 12%). Fifteen of 108 biopsies (14%) (95% CI, 7% to 21%) changed systemic therapy. Most treatment changes were for cutaneous reactions to drugs; biopsy never resulted in the diagnosis of untreated systemic infection. Biopsy findings that altered diagnoses were not more likely to change therapy. Among the 82 rashes in which biopsies were not performed, review of the chart revealed no adverse sequelae (0%) (95% CI, 0% to 5%), which would have made a biopsy advisable. Conclusions.—Skin biopsy findings often changed dermatologic diagnoses in immunocompromised cancer patients, but treatment changes based on biopsy results were much less common, and altered diagnoses in patients who underwent biopsy often did not change therapy. Untreated systemic infection was never diagnosed by means of a skin biopsy. Skin biopsies of these rashes may not be mandatory for either diagnostic or therapeutic reasons. © 1993, American Medical Association. All rights reserved.