For an object A to be gravitationally bound to another body B that is orbiting a larger body C itself, A must be inside body B's Hill sphere.
Now, the derivation of the radius of the Hill sphere does not take into account things like weirdly-shaped space stations with very complex gravitational fields, but rather it assumes perfectly spherically symmetric bodies B, C, and a massless A.
So the concept doesn't apply as-is, but let's use it anyway by making the assumption that all of the ISS is compressed into a small sphere of homogeneous density, as dense as its materials will allow. Taking this "idealized" ISS as an example, the following numbers apply:
- mass: roughly 450,000 kg
- altitude: between 435 km and 330 km.
- with the mean radius of Earth 6371km, this implies
- a semi-major axis of 6753.5 km
- an orbital eccentricity of 0.0078
Then, using 5.972e24 kg for the mass of the Earth, the radius of the ISS' Hill sphere is about 2 metres.
The Hill sphere is a more complete definition of the sphere of influence, which is the region in space where the gravity of body B dominates over body C. For the ISS, the sphere of influence is about 15 cm.
So given these numbers, and knowing that it's true gravitational field is vastly more complex than just that little idealized sphere, it's pretty much impossible to orbit the ISS. As AlanSE noted, you can put things in apparent orbits, but these are normally only transient and will cease being close to the ISS after a few dozen of these "orbits". Another way to understand that is by looking at the three body problem, particularly at the derivation of the Lagrange points. The thing to note is that the Hill sphere is the region where there is orbital stability (in the phase space of the differential equations that is, not celestial orbits), meaning, body's that start just outside the Hill sphere will show diverging orbital behaviour, whereas bodies that start just inside the Hill sphere will show stable or converging orbits.
Things will change though if the ISS were placed in deep interalactic space, far away from any celestial body. In principle, perturbations from all those remote sources will be completely negligible, and the ISS would gravitationally dominate a vast region in space, making orbits about it possible. Don't expect these orbits to be fast though; I haven't run the numbers, but I expect anything thrown faster than a few mm/s will already be moving beyond ISS escape velocity. Also, don't expect these orbits to be anywhere near Keplerian; as I mentioned, the mass distribution of the ISS is nowhere near regular, and so the orbits about it will also deviate significantly from nice conic sections.