Normally you buy a deoxyribonucleoside triphosphate derivative, where the phosphate in the alpha (the innermost) position is replaced by 35S-sulfate. These derivatives are then used for labelling by nick translation or multiprime synthesis or similar.
Advantages are:
* With 90 days, radioactive decay is much slower compared with the 14 days of 32P-phosphate. You can use the labelled nucleotide longer.
* The beta energy is much lower fir 35S (0.167 MeV) than for 32P (1.7 MeV). Consequently, the maximum distance of irradiation is lower. This means that, in autoradiography, you will have narrower bands and thus a higher resolution.
* You need much less protection from radiation due to the lower energy.
Disadvantages are:
* Specific radioactivity (activity per mole) is much lower due to the longer half-life. If you need highest sensitivity for your experiments 32P-labelled substances are to be preferred.
* Due to longer half-life time, disposal of waste is more expensive. It is not normally possible to wait long enough for being allowed to dispose of waste via normal garbage.
* Lower energy means also that it is not always easy to find spillage. In that respect 32P is much better.
Thank you Denis, of course you are right. Luckily, for hybridisation purposes, it does not matter, if you join the neighpouring positions by phosphate or sulfate. There is an alternative, however, if a low energy radioisotope is intended: 33P-labelled nucleotides. This works nicely, and looks better! - When I was young, I did most of my sequencing work with either 35S- (mostly) or 33P-labelling.