If you are talking about a tank-like loop used for bandpass (as in the schematic below)...
The capacitor comes into play by diverting high frequencies. The coil acts by diverting lows.
You will get some resonating action near the center frequency.
The input resistance will be a factor in determining what values you use for coil and capacitor.
For instance, if you use a 1pF capacitor, then a few ohms will yield a time constant suitable for the gigahertz range.
You can change the bandwidth by making the rolloff curves overlap to a greater or lesser extent. Shift the capacitor rolloff curve upward or downward by changing capacitor value or resistor value. In a similar manner you can shift the coil rolloff curve upward or downward. Always you must keep track of how you are altering the center frequency.
Changing the input resistor value will alter all time constants.
By making the outgoing resistance substantially higher than the input resistance, you reduce its ability to throw off the filter response.
---------- Post added at 03:26 ---------- Previous post was at 03:21 ----------
The simple filter shown above will not give you as sharp attenuation as your specs require.
The concepts above can be adapted to whatever type of filter you are using.
---------- Post added at 03:45 ---------- Previous post was at 03:26 ----------
In case you are using an LC series bandpass, per diagram...
For this arrangement the output (rightmost) resistor needs to be low enough to allow some current flow.
A very high resistance will prevent the springy action which you want to occur at the bandpass frequency.