The response of beetle and ant assemblages to fire (2–5 y old) and the dependence of that response on habitat type were analyzed. Using pitfall traps, beetles (Coleoptera) and ants (Formicidae) were sampled in replicated forest, scrub, and steppe areas including paired unburnt and burnt plots. A total of 176 species of beetles (8245 individuals) and 22 species of ants (115,056 individuals) were captured. Most beetle families (65%) were captured in the forest, while most ant species were most frequently captured in the steppe (45%). Beetle abundance was the same in unburnt and burnt plots in the three habitats, but ant abundance was higher in burnt than in unburnt plots. Fire usually decreased species richness of both taxa, but the strength of this effect depended on the habitat type. Coleoptera richness was lower in burnt than in unburnt plots in the forest and steppe but was the same in unburnt and burnt scrub plots. Ant richness was lower in burnt forest and scrub plots and was similar between burnt and unburnt steppe plots. For both taxa, species composition of the forest assemblages was different between unburnt and burnt forest plots (difference almost significant for ants) but not between unburnt and burnt steppe plots. Beetle species composition in the scrub was different between unburnt and burnt plots, but ant species composition was the same. Our results support the idea that the consequences of fire on native beetle and ant assemblages depend on the habitat type but also on the taxonomic group under analysis. Beetles and ants showed differences in their habitat preference and in the way that they use the habitats (e.g., preference for vegetated or bare soil patches) along the environmental gradient. These differences explain why beetles were in general more severely affected than ants, and why fire differentially affected both taxa in the scrub (habitat of intermediate complexity). A good knowledge of the habitat utilization by different groups at the local scale and a regional perspective (e.g., habitat preference along an environmental gradient) are necessary to fully understand the effect of disturbances on native arthropod assemblages.