Mulliken atomic charges can be defined as[1]:
$$q_A = Z_A-\sum_{\mu\in A}\left( \mathbf{P\cdot S} \right)_{\mu\mu} \tag{Szabo 3.196}$$
Here I have used the same notation as in Szabo[1], with $\mathbf{P}$ being the density matrix and $\mathbf{S}$ beign the overlap matrix. $Z$ is the nuclear charge, and using the greek letter $\mu$ as indicies indicates that we are working in atomic orbital basis (not in in molecular orbital basis).
The sum runs over $\mu\in A$ meaning, that we only consider atomic orbitals that are centered on the $A$th atom.
we can therefore note that Mulliken charges are only defined when we used atomic-centered basis functions (which we most often do).
[1] : Modern Quantum Chemistry: Introduction to Advanced Electronic Structure Theory, Attila Szabo and Neil S. Ostlund