Botulinum neurotoxin serotype C (BoNT/C) is a neuroparalytic toxin associated with outbreaks of animal botulism, particularly in birds, and is the only BoNT known to cleave two different SNARE proteins, SNAP-25 and syntaxin. BoNT/C was shown to be a good substitute for BoNT/A1 in human dystonia therapy because of its long lasting effects and absence of neuromuscular damage. Two triple mutants of BoNT/C, namely BoNT/C S51T/R52N/N53P (BoNT/C α-51) and BoNT/C L200W/M221W/I226W (BoNT/C α-3W), were recently reported to selectively cleave syntaxin and have been used here to evaluate the individual contribution of SNAP-25 and syntaxin cleavage to the effect of BoNT/C in vivo. Although BoNT/C α-51 and BoNT/C α-3W toxins cleave syntaxin with similar efficiency, we unexpectedly found also cleavage of SNAP-25, although to a lesser extent than wild type BoNT/C. Interestingly, the BoNT/C mutants exhibit reduced lethality compared to wild type toxin, a result that correlated with their residual activity against SNAP-25. In spite of this, a local injection of BoNT/C α-51 persistently impairs neuromuscular junction activity. This is due to an initial phase in which SNAP-25 cleavage causes a complete blockade of neurotransmission, and to a second phase of incomplete impairment ascribable to syntaxin cleavage. Together, these results indicate that neuroparalysis of BoNT/C at the neuromuscular junction is due to SNAP-25 cleavage, while the proteolysis of syntaxin provides a substantial, but incomplete, neuromuscular impairment. In light of this evidence, we discuss a possible clinical use of BoNT/C α-51 as a botulinum neurotoxin endowed with a wide safety margin and a long lasting effect.