Since their invention in 2009, Polar Codes have emerged as new near-capacity channel codes with a great potential of replacing turbo codes and low density parity check codes in a wide range of applications. This paper investigates the use of systematic polar codes in a distributed source coding application. A system model is proposed for the compression of distributed sources with side information at the decoder, along with a theoretical approach for estimating the ideal compression ratio. The system performance is assessed in terms of compression efficiency and bit error rate, for different code parameters and source correlation levels. Simulation results show the efficiency of systematic polar codes in distributed source coding, which makes them suitable for practical distributed compression applications such as wireless sensor networks and distributed video coding. Furthermore, despite the fact that error correction performance of polar codes improves with larger blocks, results show that for a compression rate below 0.4 (i.e. compression stronger than 2.5∶1), better error rates can be achieved with smaller blocks.