The presence of cells positive for cytokeratins or prostate-specific antigen (PSA) in bone marrow aspirates (BMAs) has been used to indicate the presence of micrometastasis. The aim of this prospective study of prostate cancer patients was to determine the presence of prostate cells in blood and BMAs and to compare them with bone marrow biopsy touch prep samples. The results indicated that there was a satisfactory concordance between circulating prostate cells (CPCs) in blood and disseminated tumor cells (DTCs) in BMAs for all Gleason scores (κ>0.50). However, neither were concordant with the presence of prostate cells in bone marrow biopsies except for high-grade tumors, Gleason 8 and 9. Phenotypic characteristics of CPCs and DTCs were identical (κ>0.9) but were different than cells detected in bone marrow biopsies (κ<0.2). The expression of matrix metalloproteinase-2 (MMP-2) in bone marrow biopsies was positively associated with the Gleason score (trend Chi-squared <0.05) and may explain the differences between the presence of DTCs and the presence of prostate cells in bone marrow biopsies. If the presence of DTCs was used to indicate micrometastatic disease, 20% of patients would be misclassified compared to micrometastasis defined as patients with a positive biopsy. This may have clinical implications for patients with low-grade tumors.