Although prostate cancer is a major cause of cancer death and morbidity in Western countries, one major hindrance in the study of the biology of prostate cancer has been the limited number of laboratory models, compared with the number of models available for other neoplasms. For a long time, only three cell lines, namely LNCaP, PC3, and DU145, were routinely used to study prostate cancer in the lab. The success rate to establish cell lines from human prostate cancer tissues is low, in the 1% range. Currently, only about 10 prostate cancer cell lines are available, and many of them do not reproduce typical features of the human disease, like androgen receptor expression or prostate specific antigen (PSA) secretion. Spontaneous models in animals (including rat and dog) are practically not convenient for research purposes. Several in vivo models were artificially established by transforming prostate cells by potent oncogenes. Other models were developed by injecting prostate cancer cell lines into the prostate (orthotopic model), the vessel, or the bones of immuno-deficient mice, to mimic localized and metastatic prostate cancer. This was successfully done with the MDA PCa2b cell line. This cell line was also used to generate an in vitro model of bone metastases by a co-culture system with osteoblasts. This model allows to study the paracrine cross-talk between the two cell compartments and the resulting molecular modifications. The objective of the present article is to review the currently available model systems of prostate cancer.