Purpose: To assess safety, pharmacokinetics, maximum tolerated dose, and preliminary efficacy of bivatuzumab mertansine. Bivatuzumab is a humanized monoclonal antibody directed against CD44v6, which previously seemed to be safe in phase I radioimmunotherapy trials, whereas the conjugated mertansine is a potent maytansine derivative. Experimental Design: Patients with incurable squamous cell carcinoma of the head and neck or esophagus were eligible. Bivatuzumab was given weekly for 3 consecutive weeks by i.v. infusion. One patient was planned to be treated at each dose tier as long as toxicity did not reach grade 2; otherwise, three patients had to be treated until dose-limiting toxicity occurred. Starting dose was 20 mg/m2 and dose was subsequently escalated in steps of 20 mg/m2. Patients without disease progression mertansineand not experiencing dose-limiting toxicity were eligible for repeated courses. Blood serum samples were taken throughout the treatment period to determine the pharmacokinetic properties of bivatuzumab mertansine and to assess the human anti-bivatuzumab mertansine antibody response. Results: Seven patients received a total of 23 weekly doses of bivatuzumab mertansine. One patient at the 100 mg/m2 and one at the 120 mg/m2 level experienced stable disease during treatment phase but also developed grade 1 skin toxicity (desquamation). One of them received a second treatment course. At the highest dose level achieved in this study (140 mg/m2), one patient developed toxic epidermal necrolysis after two infusions and died. Massive apoptosis of skin keratinocytes had occurred, whereas only symptomatic therapy for skin toxicity was available. The risk-benefit assessment of all patients treated in the total phase I program (4 clinical trials, 70 patients) turned out to be negative after consideration of this case of a toxic epidermal necrolysis and the skin-related adverse events observed in the other trials. Therefore, development of the conjugate was discontinued. Interindividual variability in pharmacokinetic variables was low and exposure to BIWI1 increased proportionally with dose. No anti-bivatuzumab mertansine reactions were observed. Conclusion: The main toxicity of bivatuzumab mertansine was directed against the skin, most probably due to CD44v6 expression in this tissue. The majority of skin reactions was reversible; however, one fatal drug-related adverse event had occurred. Clinical development was discontinued before reaching maximum tolerated dose.