Growing evidence suggests that asymptomatic and mild SARS-CoV-2 infections, together comprising >95% of all infections, may be associated with lower antibody titers than severe infections. In addition, antibody levels peak a few weeks after infection and decay gradually. Yet, positive controls used for determining the sensitivity of serological assays are usually limited to samples from hospitalized patients with severe disease, leading to what is commonly known as spectrum bias in estimating seroprevalence in the general population. We performed simulations to quantify the bias potentially introduced by the choice of positive controls used. Our results suggest that assays with imperfect sensitivity will underestimate the true seroprevalence, but this can be corrected if assay sensitivity in the general population is known. If sensitivity is determined from validation sets skewed towards those with severe or recent infections and thus higher antibody levels, corrected prevalence will still underestimate the true prevalence. Correct interpretation of SARS-CoV-2 seroprevalence studies requires quantifying the extent to which the sensitivity of assays being used varies with disease severity and over time. Optimization and validation of serological assays should involve samples from across the spectrum of severity and time since infection, and performance characteristics should be stratified by these factors.