Current and future cosmological surveys are targeting star-forming galaxies at z ∼1 with nebular emission lines. We use a state-of-the-art semi-analytical model of galaxy formation and evolution to explore the large-scale environment of star-forming emission line galaxies (ELGs). Model ELGs are selected such that they can be compared directly with the DEEP2, VVDS, eBOSS-SGC, and DESI surveys. The large-scale environment of the ELGs is classified using velocity-shear-tensor and tidal-tensor algorithms. Half of the model ELGs live in filaments and about a third in sheets. Model ELGs that reside in knots have the largest satellite fractions. We find that the shape of the mean halo occupation distribution of model ELGs varies widely for different large-scale environments. To interpret our results, we also study fixed number density samples of ELGs and galaxies selected using simpler criteria, with single cuts in stellar mass, star formation rate, and [O ii] luminosity. The fixed number density ELG selection produces samples that are close to L[O ii] and SFR-selected samples for densities above 10-4.2 h3 Mpc-3. ELGs with an extra cut in stellar mass applied to fix their number density, present differences in sheets and knots with respect to the other samples. ELGs, SFR, and L[O ii] selected samples with equal number density have similar large-scale bias but their clustering below separations of 1h-1 Mpc is different.