The effect of screws and pegs on the initial fixation stability of an uncemented unicondylar knee replacement.
 Two uncemented unicompartmental tibial components were examined for initial fixation stability.
 A conventional design that employed a single posteriorly angled peg was compared with a new design that was held in place by cancellous bone screws.
 The components were implanted into the medial condyles of 12 preserved human tibiae, and a cyclic load was first applied anteromedially and then posteromedially.
 The screwed implants failed at significantly higher loads (1634.8 +/- 121.6 N, mean +/- standard error of the mean) than the pegged implants (1103.3 +/- 152.0 N).
 On application of a 19.6-N preload, the screwed implants moved significantly less than the pegged implants.
 Although the differences in micromotion and subsidence were not always significant, there were definite trends.
 The screwed implants had much lower levels of temporary and permanent displacement compared with the pegged implants for all load levels from the initial load of 245.2 N up to and including the failure load.
 When the motion that resulted from moving the load from the anterior position to the posterior position was examined, the screwed implant's average total motion was less than 10 microns compared with almost 135 microns for the pegged implant after the 245.2-N load cycle.
 For the cycle before failure, the screwed implant's average motion increased to less than 29 microns, whereas the pegged implant's average total motion was almost 354 microns.
 From this information it appears clear that screws provide better initial fixation stability than angled pegs for uncemented unicondylar tibial components.
