Induction.
First, $|S_1| = 1$, trivially.
Now suppose $n > 1$. Let's count the number of permutations of the numbers $1, \ldots, n$ by counting those that send $n$ to $i$ for each $i = 1, 2, \ldots, n$. We'll show that these counts will all turn out to be the same number $c$, so the total will be $nc$, and that the number $c$ is exactly $|S_{n-1}|$, which is, by the inductive hypothesis, $(n-1)!$.
Here are the details:
Let $V_i \subset S_n$ be the set of permutations that send $n$ to $i$ for some particular $i$. To compute $|V_i|$, let $u$ be some such permutation. To each element $v$ of $V_i$, associate the element $u^{-1} \circ v$, which is an element of $V_n$. This association is clearly a bijection. So $|V_i| = |V_n|$. Thus we need only compute $|V_n|$.
Now $V_n$ consists of all permutations that send $n$ to $n$; by restricting the domain (removing $n$), we get a permutation of $1\ldots (n-1)$; this correspondence with $S_{n-1}$ is clearly invertible: to each permutation $s \in S_{n-1}$, assign the permutation
$$
s'(k) = \begin{cases} s(k) & k = 1, 2, \ldots, n-1 \\ n & k = n \end{cases}
$$
Then $s'$ is an element of $V_n$ that corresponds to $s$ under our "restriction" map.
So $|V_n| = |S_{n-1}|$.
That's it!.