Loops between low resistance components (like inductors and transformer windings) are best avoided. In the ideal case, where DCR = 0, the current in that loop is independent of the rest of the circuit, and in a SPICE simulation, you usually get a "singular matrix" error because it doesn't have any way to solve for that unknown current.
So, you're better off using two 33uH chokes in series.
Another point of note: if the inductors are not well shielded, the value obtained may be quite a bit different from the expected value (in series or parallel), due to coupling.
Tim