Twenty-one cemented total hip arthroplasties were each revised three t
imes with a new cemented prosthesis. The mean intervals between the op
erations were 54, 42 and 67 months and the mean duration of the patien
ts' follow-up was 55 months. Thus, the speed of loosening did not incr
ease from one operation to the next. At the time of follow-up, 14 pros
theses were still well fixed. The improvement in clinical results rega
rding pain and function after the third revision was comparable with t
hat achieved after the first revision. Loss of periprosthetic bone sub
stance was the major complication after multiple revisions.