In this study, symmetrical and dichotomous fractal flow channel designs with various configurations in terms of the number of channels being split into smaller channels, were investigated experimentally for their flow boiling heat transfer performance with FC-72 as the working fluid. A multi-channel parallel channel design depicted as Parallel was used as a benchmark. The channels were fabricated by Selective Laser Melting (SLM). Three mass fluxes were studied, viz., 600, 900 and 1200 kg/m2⋅s. The peak heat transfer coefficient and pressure drop were investigated. High-speed visualization studies were also employed to examine the two-phase behaviors in the channels. The fractal flow channels were found to perform comparably with the Parallel channel at low mass fluxes but achieved higher heat transfer coefficients at higher mass fluxes, with the design n = 2, i.e., channel splitting into smaller channels twice, being the highest. The pressure drops of the fractal flow channel designs are higher than the Parallel channel due to their longer channel lengths, and the flow impacting multiple bifurcations at an angle, with n = 4, i.e., channel splitting into smaller channels four times, experiencing the highest pressure drop for all mass fluxes. Visualization studies show that fractal flow channels cause an early annular flow, enhancing heat transfer. Flow reversals in the Parallel channel were observed especially at higher mass fluxes, which suggest no enhancement in the heat transfer coefficient. The fractal flow channels, however, do not experience any flow reversal which could be attributed to the acceleration of the two-phase flow after each bifurcation, which in turn increases the inertia force to overcome the force from vapor expansion. The visualization studies also suggest that increasing the complexity does not result in better flow boiling heat transfer performance. This is due to the multiple flow separations which resulted in no two-phase heat transfer interaction and thus, under-utilization of the entire channel surface area for heat transfer. A comparison with a general correlation for flow boiling gives mean absolute errors of 31.6% and 35.3% for the fractal flow channels and Parallel, respectively.