The errors contained in slant total electron content (STEC) have a strong impact on the image generated by ionosphere tomography. This paper presents a method that rejects abnormal corrections and rays (RACR) in the multiplicative algebraic reconstruction technique (MART) algorithm by applying a correction threshold and a rejecting ratio threshold. The RACR algorithm was validated using ionosonde observations, Swarm satellite measurements, independent STEC observations and a vertical total electron content (TEC) map. Its performance was compared with the MART algorithm on both geomagnetically quiet days and disturbed days. The results show that the RACA algorithm is able to capture the main phase and the recovery phase of a storm and outperforms the MART algorithm under both geomagnetic conditions. The average improvements over the MART algorithm are 36.01%, 36.56%, 6.18%, 22.10% and 6.03% in the validation tests of the peak density of F2 layer, peak height of F2 layer, the electron density of the topside ionosphere, STEC and VTEC, respectively. The quality of the image produced by the RACR algorithm was controlled by the correction threshold and the rejection threshold. Smaller threshold values tend to make the image smoother. The RACR algorithm provides not only a way to produce a better tomographic image but also a means to detect abnormal rays.