We have intensely followed 45 consecutive women who underwent high-dose chemotherapy (cyclophosphamide/cisplatin/BCNU) and autologous bone marrow transplant (HDC/ABMT) for primary breast cancer with pulmonary function testing and computed tomography at regular intervals up to 126 wk (median follow-up, 72 wk). Our results show a high incidence of interstitial pneumonitis requiring steroids (64%), but no deaths due to pulmonary toxicity. The DL(CO) reaches a nadir of 58.2 +/- SEM 3.4 (expressed as a percent of baseline value) 15-18 wk following HDC/ABMT, and marginally improves with time. To a much lesser extent, vital capacity is reduced with a parallel drop in FEV1, suggesting mild restrictive changes without significant obstruction. Patients who develop pulmonary symptoms of cough or dyspnea have a corresponding significantly greater and earlier decline in DL(CO). Chest computed tomography was neither sensitive nor specific for diagnosing pulmonary toxicity. For patients who received steroids for pulmonary toxicity, there was a subsequent improvement in DL(CO) of 17.1% (p = 0.0001). Because our patients do not fit with the recent definition of idiopathic pulmonary syndrome (IPS), we propose the term delayed pulmonary toxicity syndrome (DPTS) to better describe the milder form of lung toxicity seen in our patient population. We were unable to correlate the severity of DPTS with age, tobacco use, baseline pulmonary function, or systemic exposure to BCNU, cyclophosphamide, or cisplatin. These data suggest that factor(s) other than, or in addition to, chemotherapy systemic exposure can contribute to DPTS. Furthermore, early identification and institution of systemic corticosteroids may improve lung function.