In conceiving and designing a national or regional survey, difficult choices often have to be made between the competing claims of alternative modes of data collection. Would postal or internet methods be appropriate? Would telephone interviewing be more cost effective (rather than just cheaper) than face-to-face interviewing for that project? Might a mixed mode approach be feasible and desirable? As potentially new modes of data collection such as the internet become more widely available, these choices become ever more complex. But if such choices are tricky in national and regional studies, they are fiendish in multinational studies. In the first place, the choices multiply. Should each country use the same centrally-specified mode, as in the ESS to date? Or should the choice be limited to, say, two modes, as in the ISSP to date? Or, on the contrary, should a thousand flowers be allowed to bloom, with mixed mode and single mode alternatives varying by and within different countries? We know that different modes of data collection tend to generate different response rates. We also know that they sometimes generate different answers to the same (or seemingly equivalent) questions. It remains true that certain modes, such as telephone and the internet, are by no means universally (or even yet widely) available in certain countries, and that self-completion modes continue to pose serious problems for countries with low levels of literacy. So, do such differences make face-to-face interviewing the only serious option for a cross-national survey comprising a range of disparate countries? Or could different modes in different countries safely co-exist, their most obvious variances and biases having been mitigated and/or corrected for? A further dilemma faces existing multinational time series such as the ESS or the ISSP. If they are to contemplate changing or expanding their choice of modes in future rounds, how much damage might this inflict on the comparability of their measurements over time?