Face-to-face interviews are still the standard in conducting cross-national surveys. Although web surveys have many advantages, so far they have rarely been used in cross-national surveys. The main problem of using web in cross-national surveys are coverage error of people without internet access and problems with the availability of sampling frames. This study reports on a large-scale experiment with a push-to-web survey design in Croatia, Germany and Portugal to overcome these problems.We experimentally assigned individuals to a face-to-face only condition or a push-to-web condition, in which non-respondents to the web-phase of the study were followed-up by face-to-face interviewers. We additionally conducted three within-country experiments to better understand how incentive structures (in Germany) and the spacing of reminders (In Croatia) affected response rates and nonresponse bias. In Portugal, we test different within-household selection procedures.We find that in Germany and Croatia the push-to-web design was equally or more successful than the face-to-face survey in terms of the response rate and nonresponse bias. In Portugal, the push-to-web design was not successful, leading to low response rates and problems in the respondent selection process. We also find that a mix of unconditional and conditional incentives works best, and that weekly reminders work better than two-weekly reminders. Overall, we conclude that it is possible to use a push-to-web design as long as a sampling frame of individuals is available.