For finitely many independent real-valued random variables, if their maximum follows a subexponential distribution, then the tail probabilities of their sum and maximum are asymptotically equivalent.