In the senior author's 27-year experience in cementing femoral components, stems with smooth and rough surfaces were implanted. Prospective data have been kept on all patients allowing extensive long term followup. The polished Charnley stem was implanted in 168 hips from 1971 to 1975. To date, only four (2.4%) stems have required revision for aseptic loosening and four others for fracture. None showed significant bone lysis. The experience with the T-28 and TR-28 allows comparison of stems of similar geometry, but with different surface finishes. From 1972 to 1977, 209 polished T-28 stems were inserted and to date 18 (14 for fracture and four for loosening, 1.9%) have required revision. None had associated bone lysis. From 1977 to 1982, 227 second generation TR-28 matte surface finish stems (Ra 30) were implanted. Five required revision, three (2.2%) with major bone lysis. Radiographic review revealed minimal lysis in four (1.9%) polished T-28 stems and in 11 (4.9%) TR-28 stems, three of which showed major bone destruction. From 1980 to 1993, 1061 Iowa stems were implanted. In 1986 proximal precoat was added, which required additional roughening of the surface. In 1995 a 1.5% early loosening rate was reported with significant bone lysis, complicating revision. By 1998 34 (3.2%) stems have required revision, all associated with significant bone loss. Revision of one of 12 original Iowa (Ra 30) and 12 of 22 (Ra 80) grit blasted stems was needed before 5 years postoperatively. These early failures prompted the author to return to a polished stem with a geometry almost identical to the Charnley stem.