I agree. I know there are strong opinions about this but I think the balance leads first is safer for two reasons:
1) As Julez says, if the packs differ substantially in voltage, then the significant resistance of the balance leads will limit the current flow to reasonable values. You can do the numbers based on various pack imbalances and reasonable values of balance lead and contact resistance (total of around 30-40mOhms in most cases) to get the likely max currents.
2) More importantly, if you are not using a sophisticated fused parallel board like the FMA or the latest EPBuddy Paraboard, the balance leads will act as a fuse if you do something really stupid - like plug a 3S and a 4S into the parallel board. The short of one cell through the balance lead will immediately let you know something is wrong and the resistance will help protect that cell while you disconnect it real quick. The balance lead may need replacing but the pack won't.
Guess how I discovered this.
Two mains leads connecting a 3S and 4S without a balance lead connected is a lot less forgiving.
If the situation is normal - that is packs at similar voltages - then it has been unequivocally shown (by measurement rather than theorising) that there are no damaging currents. Consequently I connect balance leads first as a risk management decision.
Originally Posted by Julez
Hm. I do it the other way round. This way, the high resistance of the balance leads will prevent high currents should the packs be unevenly charged.