We present a scheme for magic state distillation using punctured polar codes. Our results build on some recent work by Bardet et al. [1] who discovered that polar codes can be described algebraically as decreasing monomial codes. Using this powerful framework, we construct tri-orthogonal codes [2] that can be used to distill magic states for the T gate. An advantage of these codes is that they permit the use of the successive cancellation decoder whose time complexity scales as O(N log(N)). We supplement this with numerical simulations for the erasure channel and dephasing channel. We obtain estimates for the dimensions and error rates for the resulting codes for block sizes up to 2 20 for the erasure channel and 2 16 for the dephasing channel. The dimension of the triply-even codes we obtain is shown to scale like O(N 0.8) for the binary erasure channel at noise rate 0.01 and O(N 0.84) for the dephasing channel at noise rate 0.001. The corresponding bit error rates drop to roughly 8 × 10 −28 for the erasure channel and 7 × 10 −15 for the dephasing channel respectively.