Survey research seems caught between telephone surveys that work less well than in the past, and internet surveys that do not yet work well enough to replace them. The solution proposed by many is to use more than one survey mode to conduct specific surveys. When a second or third mode of data collection is used to achieve certain objectives, such as improvements in coverage or the reduction of non response error, a potential consequence is to introduce measurement differences between modes. These differences may negate the improvements in coverage and response. In addition, although protocols are well developed for the administration of single mode surveys, much remains to be learned on how to link the use of different data collection modes in order to improve survey quality. In this presentation Dr. Dillman will discuss challenges of combining survey modes, using recent research he has done on the use of multiple modes in the same survey. He will also discuss his research that shows offering respondents a choice of survey mode tends to reduce overall response. Directions for future research will also be discussed regarding these difficult but essential challenges facing survey methodologists throughout the world.