Yes, you'll need to render the scene once from the point of view each shadow-casting light source. Actually point lights require you to render the scene 6 times per light, at least if you're using the standard cubemap method.

As a side note...when you have many shadow casters, you'll want to make sure you're only rendering the objects that are going to cast or receive shadows from that particular light source. You'll also want to make sure that a light's entire volume is visible on screen, so that you're not wasting time rendering shadow maps for lights that aren't even visible. You can do this via frustum culling, or some other method.

Regarding MJP's last comment, read Advanced Light and Shadow Culling Methods by Eric Lengyel from GDC'06 - it's a good coverage of this aspect of rendering.

Jack