The trimmed mean completely discards the selected percentage of extreme values, while the Winsorized mean retains the sample size by replacing them with the nearest non-extreme values.

Both the trimmed mean and the Winsorized mean are robust estimators of central tendency that reduce the influence of outliers.

For a symmetric dataset with no outliers, the trimmed mean, Winsorized mean, and arithmetic mean will all be very close or identical.

The Winsorized mean is always larger than the trimmed mean for any dataset containing outliers.