Not a picross master or anything: whenever I end up with rows that are clearly faulty, I just wipe them out. Often, one row is faulty due to another row or sequence of rows that are orthogonal to it; I have found that just removing a single row will often not result in fixing an error that is more deeply embedded in your puzzle. It helps to *always* go back to other rows or columns that you aren't positive about, and move them around a bit. It goes without saying, but your "core" rows (I don't have a better name for this; just the rows that you are 100% sure are correct) should be used as the guiding light you make all decisions from.
Sometimes you'll get focused on a row that you have solved, and be trying desperately to make another row adjacent it to be solved: if it's giving you trouble, the rows to wipe out are the ones you love the most: you're almost certainly "too close" to the problem.
At the end of the day, it doesn't hurt to wipe out a lot of what you've done; anything that gets you closer to a finished solution is more important. I haven't *completely* wiped out a puzzle yet, but there are some where I've knocked out a quadrant or two that seemed too difficult to be correct.
Last but not least: never focus on just 1 row / column. Always be cognizant that you need to solve a row in terms of others, usually; go slower so you can sure you're laying down a solution for a row that solves one or more additional rows. When you start to see very few rows and columns that are left, you know you're close to a solution, but that doesn't mean that you won't have to knock a quadrant loose; it's not a big deal, though, because with 3/4 of the puzzle more-or-less 'solved', it becomes much easier and faster to test ideas / solutions.