A method for the determination of total arsenic (As) in human urine by hydride generation atomic fluorescence spectrometry (HG-AFS) is described. Total As includes the four toxic species; arsenous acid, i.e., As(III), arsenic acid, i.e., As(V), monomethylarsonic acid (MMA), and dimethylarsinic acid (DMA) and the three major dietary species trimethylarsine oxide (TMAO), arsenobetaine (AsB), and arsenocholine (AsC). It is necessary to normalize signals from all As compounds present beforehand because the HG reaction is dependent on the species. We evaluated three on-line oxidation systems designed to convert As into As(V): (1) microwave-assisted heating (MW); (2) UV-photooxidation (UV-1); and (3) UV-photooxidation with post-reaction heating (UV-2). Each system was used with K₂S₂O₈/NaOH as the oxidant, and was coupled to a commercial HG-AFS instrument for As detection. The efficiencies of the three on-line oxidation systems were compared. Under optimal conditions, all three achieved a signal (normalized to As(V)) within a range of 90–110% at a given concentration, for each of the six As species studied. Calibration using aqueous As(V) standards was achieved for the two UV systems while matrix-matched calibration, i.e., base urine spiked with As(V), was mandatory for the MW system. The requirement for matrix matching with the MW is a key difference compared to the UV systems making the former less suitable for routine analyses. Method detection limits (3 SD) for MW, UV-1, and UV-2 were 9.4, 6.0, and 2.7 μg L⁻¹ respectively, detected as As(V). The performance of each system was evaluated by analysis of certified reference materials and archived urine proficiency testing (PT) materials from three different External Quality Assessment (EQA) schemes. Values obtained for archived PT materials with all three methods were within the acceptable ranges established by each EQA scheme. However, the UV-2 arrangement had some advantages with respect to a lower detection limit and shorter analysis times compared to the other systems.