Background Several studies have reported a low prevalence of current smoking among hospitalized COVID-19 cases; however, no definitive conclusions can be drawn. Objective We investigated the association of tobacco smoke exposure with nasopharyngeal swab (NPS) test results for SARS-CoV-2 infection and disease severity accounting for possible confounders. Methods The nationwide, self-administered, cross-sectional web-based Italian National Epidemiological Survey on COVID-19 (EPICOVID19) was administered to an Italian population of 198,822 adult volunteers who filled in an online questionnaire between April 13 and June 2, 2020. For this study, we analyzed 6857 individuals with known NPS test results. The associations of smoking status and the dose-response relationship with a positive NPS test result and infection severity were calculated as odds ratios (ORs) with 95% CIs by means of logistic and multinomial regression models adjusting for sociodemographic, clinical, and behavioral characteristics. Results Out of the 6857 individuals (mean age 47.9 years, SD 14.1; 4516/6857, 65.9% female), 63.2% (4334/6857) had never smoked, 21.3% (1463/6857) were former smokers, and 15.5% (1060/6857) were current smokers. Compared to nonsmokers, current smokers were younger, were more educated, were less affected by chronic diseases, reported COVID-19–like symptoms less frequently, were less frequently hospitalized, and less frequently tested positive for COVID-19. In multivariate analysis, current smokers had almost half the odds of a positive NPS test result (OR 0.54, 95% CI 0.45-0.65) compared to nonsmokers. We also found a dose-dependent relationship with tobacco smoke: mild smokers (adjusted OR [aOR] 0.76, 95% CI 0.55-1.05), moderate smokers (aOR 0.56, 95% CI 0.42-0.73), and heavy smokers (aOR 0.38, 95% CI 0.27-0.53). This inverse association also persisted when considering the severity of the infection. Current smokers had a statistically significantly lower probability of having asymptomatic (aOR 0.50, 95% CI 0.27-0.92), mild (aOR 0.65, 95% CI 0.53-0.81), and severe infections (aOR 0.27, 95% CI 0.17-0.42) compared to those who never smoked. Conclusions Current smoking was negatively associated with SARS-CoV-2 infection with a dose-dependent relationship. Ad hoc experimental studies are needed to elucidate the mechanisms underlying this association. Trial Registration ClinicalTrials.gov NCT04471701; https://clinicaltrials.gov/ct2/show/NCT04471701