Phantom studies were conducted to investigate the impact of trabecular structure on the T2' signal measured by MRI. For a separation of density from structural effects, several phantoms were built. They consisted of parallel polyethylene strings arranged in a variety of different patterns to simulate a) a constant uniform trabecular distribution with increasing trabecular thickness and b) different structures with identical overall trabecular density. An asymmetric spin echo sequence was used to determine the apparent relaxation time T2'. Changes in T2' are induced by susceptibility differences between the polyethylene strings simulating trabeculae and Gd-DTPA doped saline simulating bone marrow. The results showed an increasing T2' decay rate with a) decreasing spacing while the string density was constant and b) with increasing string density while the string arrangement was constant. The results demonstrate that the T2' signal is affected not only by density but also by spatial distribution. However, the results also indicate that a separation of the two effects is not possible from a T2' measurement alone, but that e.g., a matching CT slice that would provide purely density information is additionally needed. Theoretical simulations confirmed these results.