The authors comment on the paper by P. Myszkorowski published in Automatica, 1994, Vol. 30, pp. 913-914. In that paper, the author proposes a new sufficient condition for stability of discrete-time linear systems $x_{k+1}=A(k)x_k$, where for every $k$, components $a_{ij}(k)$ of the matrix $A(k)$ belong to the known intervals $[a^-_{ij},a^+_{ij}]$. Myszkorowski's criterion is difficult to check. The authors show that his criterion is equivalent to the easily checkable fact that $I-B$ is an $M-$matrix, where $I$ is a unit matrix, $b_{ij}=\max(|a^-_{jj}|,|a^+_{ij}|)$, and an $M-$matrix is a matrix with non-positive off-diagonal entries for which successive leading principal minors are all positive.