Resistance to raltegravir (RAL), the first HIV-1 integrase (IN) inhibitor approved by the FDA, involves three genetic pathways: IN mutations N155H, Q148H/R/K, and Y143H/R/C. Those mutations are generally associated with secondary point mutations. The resulting mutant viruses show a high degree of resistance against RAL but somehow are affected in their replication capacity. Clinical and virological data indicate the high relevance of the combination G140S + Q148H because of its limited impact on HIV replication and very high resistance to RAL. Here, we report how mutations at the amino acid residues 140, 148, and 155 affect IN enzymatic activity and RAL resistance. We show that single mutations at position 140 have limited impact on 3'-processing (3'-P) but severely inactivate strand transfer (ST). On the other hand, single mutations at position 148 have a more profound effect and inactivate both 3'-P and ST. By examining systematically all of the double mutants at the 140 and 148 positions, we demonstrate that only the combination G140S + Q148H is able to restore the catalytic properties of IN. This rescue only operates in cis when both the 140S and 148H mutations are in the same IN polypeptide flexible loop. Finally, we show that the G140S-Q148H double mutant exhibits the highest resistance to RAL. It also confers cross-resistance to elvitegravir but less to G-quadraduplex inhibitors such as zintevir. Our results demonstrate that IN mutations at positions 140 and 148 in the IN flexible loop can account for the phenotype of RAL-resistant viruses.