NSD has been member of the International Social Survey Programme (ISSP) since 1989. From the beginning, the ISSP surveys in Norway were administered as self-completion, paper and pencil mail surveys, the target population being people aged 16/18 to 79 years, residing in Norway, regardless of  nationality and dwelling. The survey samples have been selected as SRS from the Central Register of Persons in Norway, providing sampling lists of named individuals with addresses and the characteristics gender, year of birth and marital status. Up to 2004 the unit nonresponse was stable, varying between 37 and 43 percent. With respect to gender and age net samples were slightly biased, due to higher unit nonresponse among elderly women (65+) and younger men (34-), beyond what could be ascribed to random variation only. From 2004 we have however seen the well-known picture of dramatically increasing unit nonresponse, reaching 50 percent in 2005 and 57 percent in 2007. Since 2008 it has levelled out on approximately 60 percent. The increased unit nonresponse has not been evenly distributed by gender and age. Younger rather than elderly women now seem to be underrepresented in the net samples, but this is not yet a consistent trend. Nonresponse among younger men has however increased more than other age-by-gender groups, reaching as high as 74 percent. High nonresponse rates pose a major threat to survey quality as they might cause biased estimators for the target population when respondents differ from nonrespondents on the variables of interest. As seen above, significant bias and changing bias patterns can be identified with respect to age and gender. The same could apply to important characteristics we can’t control for directly on individual level, for example education. During the last years NSD has therefore increasingly focused on planned mixed-mode designs, based on assumptions emerging in the literature about than ten years ago, that tailored questionnaires and implementations offering respondents (some) freedom of choice can reduce nonresponse and also improve data quality. In the ISSP survey fielded early in 2011, a web and a paper questionnaire were offered to the sample units A split-ballot design testing contact design (first contact and further follow-up of respondents) was implemented. In the presentation we will report on some results from the survey with respect to nonresponse, mode preferences, split ballot and data quality.