52 Background: Team-based cancer care models promote primary care providers taking on a larger role in survivorship care. However, little is known about PCP attitudes toward and propensity to order non-recommended surveillance testing. Methods: A stratified random sample of PCPs identified by early-stage breast cancer patients diagnosed in 2013-15 who participated in iCanCare Study (Georgia and Los Angeles SEER registries) were surveyed about their experiences caring for cancer patients (N = 519, 58% response rate). PCPs were asked in a clinical vignette whether they would order non-recommended bone scans, other imaging (i.e. PET) or tumor marker testing (i.e. CA-125) in an asymptomatic, early-stage breast cancer survivor. A composite score was created by averaging their responses to the individual items, and categorized by tertiles into low, selective and high propensity to order non-recommended testing. PCP confidence in their knowledge about appropriate testing was also measured (5 pt. Likert-type scale; not all all-very confident) and compared with propensity to order. Multivariable, weighted, multinomial logistic regression was used to then evaluate PCP-reported factors associated with high and selective propensity to order non-recommended testing (vs. low). Results: In this sample, 32% of PCPs had a low propensity, 40% selective, and 28% a high propensity to order non-recommended surveillance tests. Of the 80% of PCPs who reported they were confident in their knowledge about appropriate testing, 27% had a high propensity to order. PCPs practicing in staff-model HMOs were less likely to have a high or selective propensity to order (vs. low) when compared to PCPs in private practice. (high aOR: 0.4, 95%CI: 0.2-0.7; selective aOR: 0.3, 95%CI: 0.2-0.6). Conclusions: Over a quarter of PCPs had a high propensity to order non-recommended surveillance testing for early-stage breast cancer patients, yet the majority reported they were confident in their knowledge about appropriate surveillance testing. Efforts to increase PCP knowledge about the specifics of breast cancer surveillance may be warranted to reduce the overuse of non-recommended testing during survivorship.