Purpose: To quantify the relative motion of the pelvic lymph nodes (LNs), seminal vesicles (SV) and prostate and define indicative margins for image-guided radiotherapy based on bony anatomy or prostate correction strategies for a 3 or 6 degrees-of-freedom couch. Methods and Materials: Nineteen patients had a planning computed tomography (CT) scan followed by a mean of 11 repeated CT scans during radiation therapy. The prostate, SV, and external and internal iliac LN regions on the left and right were outlined on each CT scan. Systematic and random uncertainties were determined along with correlations between the motions of these regions. The clinical target volume to planning target volume margins required to take only motion into account were calculated for each guidance method. Results: For bone guidance, motion of the prostate and LNs was largely uncorrelated. Margins to compensate for motion (left-right, superior-inferior, anterior-posterior, in cm) based on a 3-DOF couch were as follows: prostate (0.2, 0.6, 0.8), SV (0.4, 0.9, 1.0), and LNs (0.3, 0.4, 0.6). For prostate guidance, margins were calculated for correlated motion: prostate (0, 0, 0), SV (0.3, 0.5, 0.4), and LNs (0.3, 0.5, 0.9). For a 6-DOF couch, these margins were as follows: prostate (0.2, 0.6, 0.8), SV (0.3, 0.9, 1.0), and LNs (0.3, 0.4, 0.3) for bone guidance. For prostate guidance, margins were as follows: prostate (0, 0, 0), SV (0.2, 0.5, 0.4), and LNs (0.3, 0.6, 0.6). Conclusions: Image guided radiation therapy based on bony anatomy requires larger prostate and SV margins, and guidance on prostate requires larger LN margins. Neither guidance strategy is optimal, and a combination of the 2 or treatment adaptation after a number of fractions might be preferable. Calculation of the total margin should also include delineation uncertainties.