\left and \right are delimiter counterparts and TeX primitives, to be used with so-called "extensible delimiters" in math mode within the same group. This implies two things:
- If you use one, you have to use the other as well; and
- They have to be opened/closed at and within the same group level depth.
Common or typical uses include
\left\{ ... \right\}
\left[ ... \right]
\left( ... \right)
although there is no requirement for the delimiters to be matching. For example, it is fine to use \left\}...\right(, just as long as you use both \left and \right. For example (to mimic amsmath's bmatrix environment),

\documentclass{article}
\usepackage{amsmath}% http://ctan.org/pkg/amsmath
\begin{document}
\[
\begin{bmatrix}
1 & 2 & 3 \\
4 & 5 & 6 \\
7 & 8 & 9
\end{bmatrix} \qquad
\left[\begin{array}{@{}ccc@{}}
1 & 2 & 3 \\
4 & 5 & 6 \\
7 & 8 & 9
\end{array}\right]
\]
\end{document}
If you want to use only one of the two, you still need to include the counterpart, but this time with the argument .. That is, either \left. or \right., referred to as the null delimiter. TeX inserts a \nulldelimiterspace for these.
Extensible delimiters are ones that are comprised of a number of fixed and some variable-length parts. For some eye-candy, here's a visual of an extensible }:
