The activity of mitomycin in non-small cell lung cancer (NSCLC) has been well documented in both single-institution pilot and multi-institution randomized trials. Despite the inclusion of patients ineligible for current trials of newer single agents due to poor performance status, prior irradiation to indicator lesions, or prior chemotherapy, mitomycin emerges from such prior studies as the most consistently active single agent currently available for NSCLC. While randomized trials in stage IV disease demonstrate an improved response rate with mitomycin and cisplatin in comparison with cisplatin alone (p = 0.03), and with mitomycin, vindesine, and cisplatin in comparison with vindesine and cisplatin alone (p = 0.003), the potential with mitomycin is most apparent with weekly bolus or infused vindesine/vinblastine, and higher-dose cisplatin (MVP) in neoadjuvant approaches to stage III disease. Indeed, four trials of neoadjuvant MVP in predominantly stage IIIA (bulky N2) disease produced a consistent 19-month median and 26-33% 3-year survival, which is to be compared with 8 months and 6%, respectively, with traditional thoracic irradiation alone. Nonetheless, prior use of mitomycin without guidelines for cumulative dose, schedule, or guidelines for use in combined-modality therapy has produced widespread frustration stemming not only from frequent antineoplastic effect, but also from frequent toxicity. The several related syndromes of mitomycin-associated thrombotic microangiopathy, ranging from hemolytic-uremic syndrome to pulmonary injury, appear avoidable through limiting the cumulative mitomycin dose to a maximum of 30 mg/m2; scheduling mitomycin at not less than 4- to 6-week intervals; perioperative use of corticosteroids and low inspired oxygen; and close patient follow-up. While dexamethasone preceding mitomycin reduces the frequency and severity of mitomycin-associated lung injury (p = 0.0005), dexamethasone premedication also reduces the response rate of MVP in NSCLC (p < 0.025).