Surveys using internet access panels have been steadily replacing telephone surveys as a mode of data collection. Using the internet offers researchers advantages in terms of reduced costs, faster data collection and in some cases better data. Internet panel companies use a variety of recruitment methods and weighting techniques to make their samples as representative of the general population as possible. This paper will seek to determine to what degree internet panels are appropriate for all types of surveys. In particular we will focus on the appropriateness of using internet panels for political polls. Using the extensive and rich data collected by Mediamark’s in-person survey and data from Survey Sampling’s internet access panel, we will explore whether individuals without internet access and those who are only casual users of the internet differ from those individuals that access the internet more frequently and are therefore more likely to be members of an internet panel. Where comparable data is available, comparisons to members Survey Sampling’s SurveySpot panel will be included. We will also explore whether post-stratification of the Internet population can reduce or minimize any potential bias in survey estimates. A variety of attitudes and behavior will be analyzed within a broad set of geographic and demographic cohorts, cohorts that are normally used to weight or balance internet panels and surveys to the general population. We will concentrate on those attitudes and behaviors that are most likely to have an impact on voting behavior and therefore on the results of online political polls.