Normally the only way that could occur is if you have a skipto value that becomes active but does not get found--and consequently you end up at the end of the questionnaire/experiment. But you say this is not always happening when a skipto is active. How predictable is it? e.g., Can you select a set of specific responses which when used will always cause the questionnaire to jump to the end? If so, could you supply the que and that sequence of responses so I can try to replicate it here? Also--are you using a responses.xls file?