Centrifugation in a density gradient is the way to go, as far as I know. something like Ficoll is probably used, I cannot imagine sucrose being very good…

If you have homogenated cells you can use differential centrifugation. It’s easy, and gives good results if you measure very exactly. If you sentrifugate the cells for 10 minutes with 1,000 g you should get pellet rich in nuclei and cellular debris. Another 20 minutes with 20,000 g and you should have pellet rich in mithocondria (and chloroplasts if you are talking about plant cells.) Then, extract the mithochondria carefully with a syringe or a vacuum-pipe, and you should have the mithocondria pretty isolated.