So my understanding is that multi-threading splits it up into multiple batches and processes them all simultaneously. If you need the import to happen in the exact order they appear on your CSV (say the third record in your CSV references the first record in your CSV) or you're very particular about internal IDs or JE numbers being in a specific order then DO NOT use multi-threading, as it will just work on whatever one in whatever chunk it can get done first. As long as you don't care about the order in which your records go in, then yes, multithread away. IME transaction imports, even multiple lines ones are fine, as long as all the lines for a single transaction are adjacent to each other (but I'm pretty sure CSV import bombs out when that's not the case even for non-multithreading)