Concurrent sexual partnerships, or sexual partnerships that overlap in time, have been associated with HIV and sexually transmitted infections (STI). How best to measure concurrency and the personal characteristics and predictors of concurrency are not yet well understood. We compared two frequently used concurrency definitions, including a self-reported measure based on participant response regarding overlapping sex with partners, and the UNAIDS measure based on overlapping dates of last sex and intention to have sex again. We performed multivariable logistic regression analyses to identify socio-demographic, behavioral, and structural predictors of concurrency among 1,542 patients at an urban STI clinic in Jackson, Mississippi. Nearly half (44 %) reported concurrency based on self-reported sex with other partners, and 26 % reported concurrency according to the UNAIDS concurrency measure. Using the self-reported concurrency measure, the strongest predictors of concurrency were perceived partner concurrency, drug use at last sex, having more than 10 lifetime partners, and being recently incarcerated. Strongest predictors of concurrency using the UNAIDS measure were lifetime number of partners and perceived partner concurrency. Concurrency is highly prevalent in this population in the Deep South and social, structural and behavioral factors were important predictors of concurrency for both measures. Future research should use time anchored data collection methods and biomarkers to assess whether both definitions of concurrency are associated with HIV outcomes.