The antiviral component of Paxlovid, nirmatrelvir (NIR), forms a covalent bond with Cys145 of SARS-CoV-2 nsp5. To explore NIR resistance we designed mutations to impair binding of NIR over substrate. Using 12 Omicron (BA.1) and WA.1 SARS-CoV-2 replicons, cell-based complementation and enzymatic assays, we showed that in both strains, E166V imparted high NIR resistance (∼55-fold), with major decrease in WA1 replicon fitness (∼20-fold), but not BA.1 (∼2-fold). WA1 replicon fitness was restored by L50F. These differences may contribute to a potentially lower barrier to resistance in Omicron than WA1. E166V is rare in untreated patients, albeit more prevalent in paxlovid-treated EPIC-HR clinical trial patients. Importantly, NIR-resistant replicons with E166V or E166V/L50F remained susceptible to a) the flexible GC376, and b) PF-00835231, which forms additional interactions. Molecular dynamics simulations show steric clashes between the rigid and bulky NIR t-butyl and β-branched V166 distancing the NIR warhead from its Cys145 target. In contrast, GC376, through "wiggling and jiggling" accommodates V166 and still covalently binds Cys145. PF-00835231 uses its strategically positioned methoxy-indole to form a β-sheet and overcome E166V. Drug design based on strategic flexibility and main chain-targeting may help develop second-generation nsp5-targeting antivirals efficient against NIR-resistant viruses.