One hundred twenty-nine dogs with histologically confirmed malignant tumors were used in a prospective study to determine the toxicity of the new dihydroxyquinone derivative of anthracene, mitoxantrone, which was administered IV at 21-day intervals at dosages ranging from 2.5 to 5 mg/m2 body surface area. Each dog was evaluated for signs of toxicosis for 3 weeks after each dose was administered or until the dog died, whichever came first. The number of dogs in each evaluation period were as follows: 1 dose (n = 129), 2 doses (n = 82), 3 doses (n = 43), 4 doses (n = 26), 5 doses (n = 19), 6 doses (n = 9), 7 doses (n = 6), 8 doses (n = 5), 9 doses (n = 3), and 10 doses (n = 1). The most common signs of toxicosis were vomiting, diarrhea, anorexia, and sepsis secondary to myelosuppression. None of the dogs died of complications resulting from mitoxantrone treatment. Dogs with signs of toxicosis during the 21-day interval from administration of the first dose of mitoxantrone were 95 times (P = 0.003) more likely to develop signs of toxicosis during the 21-day interval from the second dose of mitoxantrone. Similarly, dogs that developed signs of toxicosis during the 21-day interval from the administration of the second dose were 34 times (P less than 0.001) more likely to develop signs of toxicosis during the 21-day interval from the administration of the third dose. With each 1 mg/m2 increase in mitoxantrone, the odds of developing signs of toxicosis increased by 5.9 fold (P less than 0.001).(ABSTRACT TRUNCATED AT 250 WORDS)