An iterative exponential filtering(EF) inverse projection reconstruction algorithm is proposed for data reconstruction in the field of optical tomography. The proposed algorithm combines the advantages of the high reconstruction quality of iterative reconstruction algorithm and high reconstruction speed of the EF inverse projection reconstruction algorithm. The filter function adopts the exponential function, and therefore, its antinoise performance is better than those of traditional filter functions. The algorithm adopts the normalized mean square distance d and normalized mean absolute distance r between the reconstructed and real images as the optimization goals. In addition, it adjusts the filter function exponential factor to reduce the impact of projection data noise and establishes an iterative calculation model. The simulation experiments show that the EF function has better reconstruction accuracy than traditional functions. The image reconstruction quality of the proposed algorithm is higher than that of the EF inverse projection reconstruction algorithm, r decreases by 20%. A refractive index optical tomography is conducted. The refractive index reconstruction of the measured data is performed using the proposed algorithm and EF inverse projection reconstruction algorithm. The reconstruction results are then compared with the calibration results obtained using the instrument. The results show that the proposed algorithm exhibits high reconstruction accuracy, and that the maximum error with the calibration results obtained using the instrument is 7.9×10^-6. Compared with the EF inverse projection reconstruction algorithm, the reconstruction accuracy of the proposed method is improved by approximately 21%.