Recall that a distribution is called subexponential if the decay of the survival function of the convolution product has twice the size as the original survival function as \(t\to\infty\). The class of subexponential distributions \(S\) is studied under the operations \(\max(X,Y)\) and \(\min(X,Y)\), where \(X\) and \(Y\) are independent random variables with distribution in \(S\). It is shown that \(\max(X,Y)\) is in \(S\) if and only if \(X+Y\) belongs to \(S\). The case is more complicated for \(\min(X,Y)\).