McGrath and Hollingshead's Task-Media Fit Circumflex, which proposed that the negotiations tasks “best fit” highly information-rich channels, is widely accepted for more than two decades. However, this study argues that according to the task nature, different negotiation tasks shall “best fit” different media channels. A 2 × 3 laboratory experiment was conducted to test this argument. Findings suggested that higher process satisfaction, lower cognitive effort and higher negotiation efficiency would be created when text-based systems were used to execute more analyzable tasks. In contrast, less analyzable tasks best fit face-to-face or audio/video-based systems. Managerial implications are discussed accordingly.