sciences states that the reps dont matter as much the effort put in to the last rep. so whether you are doing 5, 10, 15, or 19 reps, just be sure to push till muscle failure for the set. rinse, and repeat. this is how some people seem to gain on one rep range while others gain on other rep ranges. people can gain on any rep range as long as the effort is put in well enough. which rep range do you believe in? as that is the rep range you should train in.
long usage of periodization by strength athletes has shown that varying the reps within a cycle or from cycle to cycle gain garner even better gains. i like the idea of training like a powerlifter, do accessory work like a bodybuilder, do conditioning like a strongman.
I agree with this and would add small bit to it. I find that if you are training to failure, % of 1RM has some impact on recovery/improvement. If you are training in the 5 reps to failure range on every set, all-out, then you are going to be treading on a fine line between stimulating your muscles for growth and over-taxing your CNS. At least, this is my theory based on my own experience.
What I find is that training in the 8-10 or 8-12 rep range has very little impact on my CNS while doing 4-7 reps to failure can be hard to recover from. I'm not saying that I don't do this rep range, because I do - but I try to add reps and keep at the weight until I'm at 8-10 or 8-12 reps depending on the exercise.
So, I may have some exercises where I'm failing at just 4 or 5 reps if I'm using a new weight, and in that same workout I may be doing another exercise that I am doing 8-10 reps and I try to avoid a situation where I'm doing the entire workout in the 4-6 range. Even given the fact I may do just 3 working sets per muscle group, using such heavy weights that I'm at 4-6 reps for all of the workout seems to slow my progress.
Once I hit 10 reps, I increase the weight. This almost creates a default periodization in a way too - I have a heavy weight and 4-6 weeks later that weight is allowing me to get 10 reps, which almost simulates a de-loading phase.