Because there are currently no reliable predictors for progression of ductal carcinoma in situ (DCIS) to invasive disease, nearly all patients receive comprehensive therapy, leading to over-treatment in many cases. Few in vitro models for studying DCIS progression have been developed. We report here the successful culture and expansion of primary DCIS from surgical specimens using a conditional reprogramming protocol. Patients with percutaneous core-needle biopsy demonstrating DCIS were enrolled in a tissue banking protocol after informed consent was received. Fresh tissue was taken from lumpectomy or mastectomy specimens, mechanically and enzymatically dissociated, cultured in medium conditioned by irradiated mouse fibroblasts and supplemented with rho-associated protein kinase (ROCK) inhibitor, and characterized by immunocytochemistry. Out of 33 DCIS cases, 58% (19) were expanded for up to 2 months in culture, and 42% (14) were frozen immediately after mechanical dissociation for future growth. The cultures are almost exclusively composed of cytokeratin 8- and EpCAM-positive luminal and cytokeratin 14-, cytokeratin 5-, and p63-positive basal mammary epithelial cells, suggesting maintenance of heterogeneity in vitro. Furthermore, as assessed by luminal and basal marker expression, these cells retain their cellular identities both in the "conditionally reprogrammed" proliferative state and after conditioned media and ROCK inhibitor withdrawal. When grown to 100 % confluency, the cultures organize into luminal and basal layers as well as luminal compartments surrounded by basal cells. Primary cultures of DCIS derived directly from patient tissues can be generated and may serve as in vitro models for the study of DCIS.