Internet panels derived from probability-based sampling are attractive to researchers confronting the limitations of using more traditional modes, such as RDD telephone, to obtain nationally representative estimates (particularly of youth). However, panel conditioning and panel attrition are potential weaknesses that can compromise panel data quality. In this paper we will examine panel conditioning effects in a cross-sectional tracking study on military advertising that involved the reuse of panelists in order to obtain the desired sample sizes. Data collection for this study began in 2006 and involved quarterly fieldings for two surveys on military advertising with overlapping content. To obtain the desired sample sizes, panel eligibility criteria were negotiated with the vendor such that (a) a panelist could not participate in both surveys during the same quarterly fielding and (b) panelists who participated in one survey would have to wait 9 months before being eligible to participate in that same survey again. In this paper, data from one of the surveys will be used (n=5584) to examine differences among the following groups: (1) prior participation in same survey (n=1253, 22%); (2) prior participation in similar survey (n=225, 4%); (3) prior participation in same survey AND similar survey (n=999, 18%) and (4) no prior participation (n=3107, 56%). Data analysis will examine group differences in responses to questions on advertising awareness, brand imagery, and product endorsement. Data quality will also be examined by looking at non-differentiation, a form of survey satisficing, and constructing tests of concurrent validity. The conclusion will offer new insights regarding the boundaries of panel conditioning and provide practical recommendations to researchers who employ Internet-panels for their own efforts.