The orally administered tubulin-binding agent 1069C85 was developed with the hope of overcoming the multidrug resistance associated with existing anti-tubulin agents, such as the vinca alkaloids. A phase I study was performed using a single oral dose every 3 weeks, administered as a suspension reconstituted in 0.1% Tween 80 and 0.9% saline. The starting dose was 2.8 mg m-2, and dose doubling was permitted until the area under curve (AUC) was > or = 40% of that at the mouse LD10; thereafter, a modified Fibonacci scheme was used. The formulation proved to be unsatisfactory, resulting in inconsistent absorption. The terminal elimination half-life was prolonged (range 18-73.5 h). Sporadic central neurotoxicity was observed, which was grade 3 in one patient treated at 200 mg m-2. A revised formulation with micronized drug was more easily suspended and appeared to increase the bioavailability by a factor of 2-4. Severe central neurotoxicity, up to grade 4, was then observed at doses of 50-100 mg m-2. Unfortunately, toxicity was not predictable and one patient, with a previous history of partial intestinal obstruction, treated at 50 mg m-2, cleared the drug very slowly, possibly because of prolonged, delayed absorption. This patient died from pancytopenia and severe gastrointestinal damage. It was concluded that such unpredictable behaviour would be incompatible with safe evaluation in phase II studies; the trial was closed and further clinical development abandoned.