Although survey breakoff appears to be somewhat less likely in online panels than in usual web surveys – due, at least in part, to early panel attrition by less-committed panel respondents (see e.g., McCutcheon, Rao, and Kaminska, forthcoming) -- breakoff in online panels remains problematic. The increased interest in breakoff involving internet survey respondents has been accelerated by the relatively recent availability of paradata collection methods for web surveys (Peytchev 2009). In addition to respondent and survey design characteristics, it is now relatively easy to obtain data such as the amount of time taken per survey item (response latency), number of response changes, time of day that survey breakoff occurs, and other factors that may contributors to survey breakoff. Internet panels offer the additional advantage of accumulated information regarding respondent characteristics. This study examines data from multiple waves of the internet component of the Gallup Panel, a multi-mode, probability panel of American households. In addition to demographic characteristics and survey design factors (e.g., question complexity, topic, number of questions, survey length), the analysis will include self-reports on internet sophistication, paradata and device type to explore factors related to survey breakoff. Preliminary analysis indicates that while long-term panel members are less likely to breakoff, past behavior is an important predictor – those who have broken off in the past appear to be more likely to do so again. Also there appears to be a clear and persistent pattern with respect to response latency; as respondents approach breaking off their survey participation, they tend to slow down in their response time (increase response latency). The study will explore the potential use of such predictive models for survey breakoff in designing possible adaptive design (Groves and Heeringa 2006) interventions for internet surveys that may prove useful in delaying survey breakoff.