Distal femoral centralizers of five different designs were inserted into model femoral stems and cemented into closed-ended tubes simulating a proximal femoral canal. Specimens underwent cyclic loading from 50 to 500 lb. for 0, 1, 2, 5, and 10 million cycles. Each specimen was then sectioned transversely at multiple levels to obtain serial cross-sections, beginning at the femoral stem tip and proceeding distally so as to include the full extent of the centralizer. The area of each section occupied by a centralizer and the total amount of porosity present in the cement surrounding the centralizers were measured using an image analyzer. A dye penetrant was then applied to each section to visualize cement cracks and areas of incomplete bonding between cement and centralizers. The number, length, and location of cement cracks were catalogued for each section. No cement cracks or lack of bonding was observed at the interface between cement and centralizers. There was greater porosity in the specimens containing centralizers than in controls without centralizers (P < .05). The cement surrounding two of the centralizer designs had a significantly smaller amount of porosity than the cement surrounding the other three designs (P < .05). The number of cracks did not depend on whether a centralizer was used, the type of centralizer, or the cycling duration. In the control specimens, failure to adequately plug the centralizer receptacle hole in the stem tip resulted in very large cement voids.