Many of the methods currently used in phase I trial design date back to the 1970's. Recently, some investigators have considered whether it might be possible to perform more efficient phase I trials that minimize the number of patients treated at biologically ineffective doses, maximize the precision of recommended phase II doses, and maintain patient safety. The objective of this Workshop was to examine aspects of phase I trial design, with consideration of both standard and novel approaches. Topics included choice of starting dose level, definition of dose-limiting toxicity, development and incorporation of non-toxicity endpoints, use of alternate dose escalation schemes, and definitions of tumor response and clinical benefit, including consideration of whether these determinations from phase I studies should alter the decision to proceed to phase II drug evaluation. For each topic, this summary includes a description of the standard approach, a summary of the speaker's presentation, and commentary. The Workshop initiated discussion and reassessment that are expected to lead to testing of some of the suggested alternate approaches.