Hi Wenhui Xi, even with twinning and translational NCS, one does not expect Rfree to be nearly 60%. Rather such a high value points to the model being at least partially wrong or incomplete. Even if you have a decent model for molecular replacement, the success of structure solution depends on the quality of the data. There are several things that can go wrong in data processing, in particular in indexing, spacegroup selection, shadow masking, radiation damage appraisal and selection of useful frame range and resolution. Wrong choices may prevent or at least impede structure solution. I am one of the authors of the XDS program, and if you want I can look at your raw data with XDS (confidentially of course) and help you with processing. In that case, contact me off-list and send me a link to your raw data. Best wishes, Kay