Abstract Background Recently, PRRT2 gene mutations have been identified as a causative factor of paroxysmal kinesigenic dyskinesia (PKD). However, evidence is still lacking with respect to the genotype to phenotype correlation in PKD patients. Methods We recruited a cohort of PKD patients with or without PRRT2 mutations for the study, and followed them for 6 months to observe the response to carbamazepine treatment. Results Thirty-four participants were included in this study; 16 patients were positive for a hot-spot p.R217Pfs*8 heterozygous PRRT2 gene mutation, while the other 18 patients were negative for PRRT2 gene mutations. PRRT2 mutations were found to be associated with a younger age of onset, bilateral presence and a higher frequency of attacks. Furthermore, the follow-up study revealed that p.R217Pfs*8-positive patients showed dramatic improvement with complete abolition of dyskinetic episodes with carbamazepine treatment, while only 7 of the 18 patients without PRRT2 mutations showed a response to the antiepileptic drug. Conclusions Our study indicated that positivity for PRRT2 mutation is a predictor of younger age of onset and more frequent of attacks in PKD patients. Interestingly, the presence of PRRT2 mutations also predicted a good response to carbamazepine therapy, especially at low dose. Therefore, genetic testing shows potential clinical significance for guiding the choice of medication for individual PKD cases.