Cannabis consumption has been increasing worldwide among pregnant women. Due to the negative effects of prenatal cannabis exposure, it is necessary to develop an objective, sensitive, and specific method to determine cannabinoids use during pregnancy. In this study, we compared four different biological samples, maternal hair, meconium, umbilical cord, and placenta, for the detection of in utero cannabis exposure. The biological samples were collected from 627 mother-newborn dyads. All hair and meconium samples were analyzed, and umbilical cord and placenta if hair and/or meconium were positive for cannabinoids. Meconium and hair showed to complement each other, with an agreement between hair and meconium results of 96.7% but only 34.3% if just positive results were considered. Umbilical cord and placenta results showed a better agreement with meconium (91.3% and 92.6%, respectively) than with hair (39.1% and 34.6%, respectively). The predominant metabolites in meconium were 11-nor-carboxy-THC (THCCOOH) and 8,11-dihydroxy-THC (diOHTHC), and in umbilical cord and placenta was THCCOOH-glucuronide. Cannabidiol (CBD) and cannabinol (CBN) were detected in meconium but not in any umbilical cord or placenta. For the first time, prenatal marijuana exposure was analyzed and compared in paired hair, meconium, umbilical cord, and placental samples.Hair and meconium positivity rate was similar, but a more sensitive and specific analytical method for the hair may resolve discrepancies between the matrices. Umbilical cord and placenta may be considered suitable alternative matrices to meconium through the determination of THCCOOH-glucuronide as a biomarker of cannabis exposure.