usually, resistors 10k and below aren't significant noise sources. if the op amp gain is reasonable (between 1 and 50) it's probably nothing to worry about. there are other ways to correct the offset, depending on what you're using the op amp for, and some of these also apply to audio power amps (big op amps using discrete components in most cases). the simplest way to balance offset is to balance the input currents as you're describing. you could also use a bias source with a potentiometer and a high value resistor. unfortunately the resistor you use to isolate the bias supply from the input impedance of the op amp is also going to generate some noise. another option is to (and this is an audio only option, as you will no longer have any gain below a certain frequency, usually between 1 hz and about 30hz, depending on component selection) use a DC blocking cap between the inverting input and ground, in series with the divider resistor. there's another option, but it uses up another op amp (which actually would be good if you have an unused op amp in a dual or quad), and that's a DC servo. so you would have an AC feedback loop, and a separate DC feedback loop that goes through a second op amp being used as an integrator, and providing DC correction for the first op amp. of all of the solutions, the DC blocking cap probably introduces the least noise, and is only a few cents more expensive than the balancing of input currents method.
the bias nulling terminals on the TL071 are linked to the current mirror, and are used as in the picture below.
another couple of things to consider are using low-noise op amps and op amps with very low bias currents. FET input op amps are usually very good in both categories. the best known of these being LF353 and TL072 op amps. there are newer improved versions of these, the TL2072 being faster, quieter and having even less input bias current than it's "grandpa" the TL072.