We address the issue of single top production in association with a W boson at the Large Hadron Collider, in particular how to obtain an accurate description in the face of the top pair production background given that the two processes interfere with each other. We stress the advantages of an MC@NLO description, and find that for cuts used to isolate the signal, it makes sense to consider Wt as a well-defined production process in that the interference with t (t) over bar production is small, and the cross-section of the former is above the scale variation uncertainty associated with the latter. We also consider the case where both W t and t (t) over bar production are backgrounds to a third process (Higgs boson production followed by decay to a W boson pair), and find in this context that interference issues can also be neglected. We discuss the generalization of our results to other situations, aided by a comparison between the MC@NLO approach and a calculation of the WWb (b) over bar final state matched to a parton shower.