The COVID-19 pandemic created a natural experiment to assess air quality responses to emission reductions, yet evidence from Latin American coastal industrial cities remains scarce. This study examined how meteorological variability modulated the effects of COVID-19 restrictions on air quality in Ilo, a medium-sized coastal industrial city in southern Peru (~67,000 inhabitants). We analyzed daily concentrations of PM10, PM2.5, NO2, O3, and SO2 across six pandemic phases (January–December 2020) using multiple linear regression, variance decomposition, and Random Forest models. A health impact assessment translated PM2.5 changes into cardiovascular mortality estimates using Global Burden of Disease 2021 coefficients. Despite reduced anthropogenic activity, PM2.5 increased by 34% during early reopening (May–June: 16.9 vs. 12.6 µg/m³ baseline), whereas NO2 decreased consistently (13–19%), SO2 declined up to 65%, and O3 more than doubled (+108%) in austral winter. Variance decomposition revealed that O₃ variability was almost entirely meteorology-driven (98%), while PM2.5 and NO2 showed balanced contributions from meteorology and restrictions (~50% each). The PM2.5 increase corresponded to approximately 3 additional cardiovascular deaths per 100,000 population annually. Coastal meteorology can counteract emission reductions, generating counterintuitive air quality responses and underscoring the need for meteorological normalization in policy evaluation.