Yes I also agree. I tend to subtract the LJP offline. My standard current clamp solutions have an LJP of 13.3mV (as calculated with the Clampex Junction Potential Tool), so the values reported by the amplifier and recorded, prior to correction, are quite incorrect.
I agree with these answers. But for voltage clamp experiments or current clamp experiments, where you want to start at a similar resting membrane potential, you may want to calculate the LJP before you do your experiments, so you can adjust your holding potentials so that you're holding at a "true" -70 mV or wherever you want to hold. The Clampex Junction Potential Tool is great.
If you don't have Clampex, here's a source of information that the Clampex LJP tool is based on that is freely accessible: http://web.med.unsw.edu.au/phbsoft/LJP_Calculator.htm
If you want to go the extra step and measure your LJP, here's a place to get started: http://junctionpotential.blogspot.com/2010/10/how-do-i-measure-liquid-junction.html