Background: Virologic detection of SARS-CoV-2 through Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) has limitations for surveillance. Serologic tests can be an important complementary approach. Objective: Assess the practical performance of RT-PCR based surveillance protocols, and the extent of undetected SARS-CoV-2 transmission in Shenzhen, China. Design: Cohort study nested in a public health response. Setting: Shenzhen, China; January-May 2020. Participants: 880 PCR-negative close-contacts of confirmed COVID-19 cases and 400 residents without known exposure (main analysis). Fifty-seven PCR-positive case contacts (timing analysis). Measurements: Virological testing by RT-PCR. Measurement of anti-SARS-CoV-2 antibodies in PCR-negative contacts 2-15 weeks after initial testing using total Ab ELISA. Rates of undetected infection, performance of RT-PCR over the course of infection, and characteristics of seropositive but PCR-negative individuals were assessed. Results: The adjusted seropositivity rate for total Ab among 880 PCR-negative close-contacts was 4.1% (95%CI, 2.9% to 5.7%), significantly higher than among residents without known exposure to cases (0.0%, 95%CI, 0.0% to 1.0%). PCR-positive cases were 8.0 times (RR; 95% CI, 5.3 to 12.7) more likely to report symptoms than the PCR-negative individuals who were seropositive, but otherwise similar. RT-PCR missed 36% (95%CI, 28% to 44%) of infected close-contacts, and false negative rates appear to be highly dependent on stage of infection. Limitations: No serological data were available on PCR-positive cases. Sample size was limited, and only 20% of PCR-negative contacts met inclusion criteria. Conclusion: Even rigorous RT-PCR testing protocols may miss a significant proportion of infections, perhaps in part due to difficulties timing testing of asymptomatics for optimal sensitivity. Surveillance and control protocols relying on RT-PCR were, nevertheless, able to contain community spread in Shenzhen.