At autopsy ≥80% of prostate cancers have established macrometastases in marrow containing bone. The mechanism(s) to explain this remarkable level of bone involvement remain to be elucidated. We examined the adhesive and invasive behavior of prostate cancer cells to osteoblastic and human bone marrow endothelial cells (HBME-1) in an attempt to explain the tropism of prostate cells for bone. We found an inverse relationship between adhesion and prostate cell tumorigenicity and metastatic potential. Relative cell adhesion of P69 between cell lines was 1.74-fold (95% confidence interval [CI] = 1.15–2.64) and 1.58-fold (95% CI = 0.94–2.68) greater at 1 hr and 2 hr, respectively, than LNCaP that was essentially equivalent to C4-2 cells when using an osteoblastic cell line, D1 as the substrate. Similar results were acquired when HBME-1 were used as substratum. There was a marked increase in adhesion of the poorly tumorigenic cell line P69 as compared to the cancer cells to HBME-1. P69 adhesion was 2.78-fold (95% CI = 1.87–4.84) and 2.0-fold (95% CI = 1.43–2.80) greater at 1 hr and 2 hr, respectively when compared to LNCaP or C4-2 cells. D1 cells, a bone homing osteoblastic precursor, behaved contrary to the metastatic, bone-colonizing C4-2 cell line and bound best to other bone cells but not as well as a non-homing fetal bone marrow-derived cell line, D2. Invasion of prostate cancer cells through HBME-1 lawns was examined at 8 hr and 16 hr. In contrast to the adhesion studies, the invasion of the more aggressive C4-2 cells was 3.46-fold (95% CI = 1.18–10.17) and 2.65-fold (95% CI = 1.26–5.56) greater at 8 hr and 16 hr, respectively than LNCaP cells. Similarly, LNCaP cell invasion was 1.73-fold (95% CI = 0.69–4.37) and 2.35-fold (95% CI = 1.41–3.93) greater at 8 hr and 16 hr, respectively than P69 cells at the invasion of HBME-1 monolayers. At 8 hr, C4-2 cells had 6.0-fold (95% CI = 2.63,13.65) higher invasive potential than P69 cells. Phage display biopanning of LNCaP cells versus C4-2 cells in vitro using 4 separate techniques repeatedly identified the same peptide in support of minimal cell surface changes associated with the ability of C4-2 cells to metastasize to bone. As integrins are vital to cell adhesion and migration, we examined the integrin subunit expression in the prostate cell lines. The expression of integrin subunits is much higher in the nontumorigenic cell line, P69, whereas the differences in integrin expression between LNCaP and C4-2 are negligible. Only α2 and β5 integrin subunits increase from LNCaP to C4-2. Given that C4-2 cells spontaneously metastasize to bone in vivo and LNCaP cells do not, these studies imply that the ability of a metastatic prostate cancer cell to colonize the bone is not completely dependent upon the ability of the cancer cell to adhere to either osteoblastic cells or to the bone marrow endothelial cell lining. Therefore, the initial interaction between the bone endothelium or stroma and prostate cells is not accurately referred to as a tropic or homing response. The invasion assay results indicate that the invasive potential of the cell more accurately reflects the bone colonizing potential of a prostate cancer cell. It is likely that bidirectional paracrine interactions, subsequent to marrow adhesion, between prostate cancer cells and the bone microenvironment are what determine the successful colonization of the bone by prostate cancer cells. Further, functional changes in surface proteins that are involved in invasion are likely to occur without major changes in levels of cell surface protein expression. Functional integrin association, substratum usage and outside in signaling are more likely to predict metastatic behavior. © 2004 Wiley-Liss, Inc.