The Polish Internet Diagnose survey was a really lengthy one: it consisted of over 40 screens summing up to several hundreds of data items. Out of 1915 participants who started with the first page, 985 completed the full set, accounting to a dropout rate of 49% - quite a remarkable success for a survey which required on the average almost exactly 1 hour to complete with a standard deviation of ca. 25 minutes. One of the methods which was applied in order to decrease the expected drop out was the pausing mechanism available in the HouseLab platform used as a vehicle for the research (see the GOR08 workshop). Selected pages, placed at the "milestones" of survey progress line, offered the participant a possibility to break and then come back within maximum 24 hours. The mechanism was based on a special link, which fully restored the application context after having used the saved bookmark. The option worked even when the participant closed the browser window or resumed work on another computer. The paper analyses the effect of the pause on the dropout rate in more detail. It turned out that relatively few participants (6.0% of those who started and 4.4% of those who completed) took advantage of the option. On the other hand, ca. 70% of those who used the pause completed the survey. Although their net working time was ca. 15 longer than for those who did not use the pause, none of the users with extremely long fillout times (over 2 hours) belonged to this group. This suggests that the pause effect helps those who are more busy, but also those who are more determined to complete the survey. The latter point is especially important, because the data provided by this participant group are usually most reliable. Therefore, although the pause mechanism targets relatively few participants, we still regard it as worthwhile implementing. Although not typically present in online research platforms, it should be seriously considered by the designers of very complex surveys.