Weed species are known to evolve rapidly with their associated crops. A better understanding of the mechanisms and rates of weed evolution could aid in limiting or at least anticipating this process. Spontaneous hybridization between crops and related weed species can transfer crop genes coding for fitness-enhancing traits to wild populations, but little is known about how easily this takes place in various weed–crop complexes. We studied interspecific hybrids between wild and cultivated radishes (Raphanus raphanistrum 3 R. sativus), which often co-occur and share pollinators. To determine whether the F1 generation represents a strong barrier to subsequent introgression, we compared the fitness of wild plants and wild–crop hybrids. Two experiments were carried out in Michigan, USA, one with potted plants and the other involving four artificially established populations. In the artificial populations, we used white flower color, a dominant, crop-specific allele, to document the persistence of crop genes over time. Wild plants had yellow flowers, which is a recessive trait. F1 hybrids had lower fitness than wild plants due to lower pollen fertility, fewer seeds per plant, and delayed flowering. Despite these disadvantages, hybrids contributed substantially to each population’s gene pool. After 3 yr, frequencies of whiteflowered plants in the artificial populations ranged from 8% to 22%, demonstrating that crop genes persisted. Other studies of flower color variation in wild populations of R. raphanistrum provide circumstantial evidence for frequent crop-to-wild gene flow. We predict that, if cultivated radish is engineered to possess transgenes coding for traits such as resistance to insect herbivores, disease, herbicides, or environmental stress, these fitnessrelated crop genes will easily spread to R. raphanistrum.