The precision of parts created by microfabrication processes is limited by surface roughness. Therefore, as a means of improving surface roughness, pulsed laser micropolishing on nickel was examined numerically and experimentally. A one-dimensional finite element method model was used to estimate the melt depth and duration for single 50–300 ns laser pulses. The critical frequency was introduced to predict the effectiveness of polishing in the spatial frequency domain. A 1064 nm Nd:YAG laser with 300 ns pulses was used to experimentally investigate pulsed laser polishing on microfabricated nickel samples with microscale line features. A microfabricated sample with wide and high lines spaced apart and one with wide and high lines spaced apart were polished with 300 ns long pulses of and fluences, respectively. The critical frequency for these experimental conditions was predicted and compared with the reduction in the average surface roughness measured for samples with two different spatial frequency contents. The average surface roughness of and wavelength line features were reduced from to and from to , respectively. Four regimes of pulsed laser micropolishing are identified as a function of laser fluence for a given pulse width: (1) at low fluences no polishing occurs due to insufficient melting, (2) moderate fluences allow sufficient melt time for surface wave damping and significant smoothing occurs, (3) increasing fluence reduces smoothing, and (4) high fluences cause roughening due to large recoil pressure and ablation. Significant improvements in average surface roughness can be achieved by pulsed laser micropolishing if the dominant frequency content of the original surface features is above the critical spatial frequency for polishing.