Mixed-mode push-to-web surveys are becoming increasingly popular as a means of combining the desirable but conflicting properties of different data collection modes. Specifically, encouraging as many sample members as possible to participate by web minimises costs, while the use of interviewer-administered modes to follow-up nonrespondents can result in improved response rates. However, push-to-web methodology is still in its infancy. Push-to-web is particularly challenging when only address lists are available as a sampling frame. The article reports the results of a randomised experiment with different ways of handling a) the invitation to all household members at each address, and b) the introduction of the mixed-mode nature of the survey. Furthermore, the push-to-web methods are compared with a traditional face-to-face approach. The experiments are carried out on a national general population sample. Push-to-web is found to be viable though response rates are slightly lower than with face-to-face. Few differences in outcomes are found between the different ways of handling the complexity of the survey context, leading to clear conclusions regarding preferred methods.