Hydrogen consumption regression models were developed for the hydrotreating of various light gas oil streams derived from Canadian oil sands including virgin light gas oil (VLGO), hydrocracker light gas oil (HLGO), coker light gas oil (KLGO), and a partially hydrotreated heavy gas oil (PHTHGO) stream over commercial NiMo/ɣ‐Al2O3 in a micro‐trickle bed reactor. The experiments were designed by central composite design (CCD) and covered a wide range of temperatures (353–387 °C), pressures (8.27–10.12 MPa), and liquid hourly space velocity (LHSV) (0.7–2.3 h−1), at H2/oil ratio = 600 m3 H2/m3 oil. A composite regression model comprising of all four feed streams was also developed and tested against a new batch of experimental data. The composite model compared favourably with the experimental data. In addition, the composite model fits better than similar correlations from the literature. The effects of process conditions on hydrodesulphurization (HDS), hydrodenitrogenation (HDN), and hydrodearomatization (HDA) conversions were also studied in this work. Based on the experimental data, regression models were developed for each feedstock to obtain the optimum conditions to maximize hydrotreating conversions.