Transiting extrasolar planets constitute only a small fraction of the range of stellar systems found to display periodic, shallow dimmings in wide-field surveys employing small-aperture camera arrays. Here we present an efficient selection strategy for follow-up observations, derived from analysis of the light curves of a sample of 67 SuperWASP targets that passed the selection tests we used in earlier papers, but which have subsequently been identified either as planet hosts or as astrophysical false positives. We determine the system parameters using Markov-chain Monte Carlo analysis of the SuperWASP light curves. We use a constrained optimization of χ2 combined with a Bayesian prior based on the main-sequence mass and radius expected from the Two Micron All Sky Survey J - H colour. The Bayesian nature of the analysis allows us to quantify both the departure of the host star from the main-sequence mass-radius relation and the probability that the companion radius is less than 1.5 Jupiter radii. When augmented by direct light-curve analyses that detect binaries with unequal primary and secondary eclipses, and objects with aperture blends that are resolved by SuperWASP, we find that only 13 of the original 67 stars, including the three known planets in the sample, would qualify for follow-up. This suggests that planet discovery `hit rates' better than one-in-five should be achievable. In addition, the stellar binaries that qualify are likely to have astrophysically interesting stellar or substellar secondaries.