A web-based experiment was conducted worldwide including eight universities from four countries. An independent and comprehensive dataset considering three colour attributes (lightness, chroma and hue) and cross-term effects was accumulated for verifying the prediction performance of colour appearance models on simultaneous colour contrast effect using categorical judgement method. The small observer variation indicates that the web-based method is robust for this experiment while offering convenience and large scale. CC_Z model derived from the authors and CCW model proposed by Wu and Wardman using Hunt94 model's colour contras t adaptation functions were tested. The results show although the latter gave a good performance on lightness contrast, the predictions on hue, chroma and cross-terms were unsatisfactory. The CCZ model performed well on all of the types of contrast effects, especially the lightness and chroma contrast. Although the result was slightly scattered for the hue contrast, it still gave a good consistency.