There is a large body of evidence that the consumption of fruit and vegetables can decrease the risk of cancer. However, the link between diet and health is extremely complex. Some dietary phytochemicals seem to offer protection in an exposure-related manner and many molecular targets and signaling pathways affected by phytochemicals have been discovered. Although in vitro studies have contributed significantly to our understanding, quite a number use concentrations orders of magnitude greater than those achievable in humans or toxic to normal tissues (exemplified by toxic concentrations of indole-3-carbinol, epigallocatechin-3-gallate, curcumin, and genistein for breast cells). Such studies may produce results that are physiologically irrelevant, thus hindering predictions of efficacy. Here, we argue for careful consideration to be given to the in vitro experimental conditions under which dietary phytochemicals are investigated. Design features, such as the use of appropriate nontoxic concentrations, extended treatment times, three-dimensional cultures, primary tumor cultures, and comparison of susceptibility of various cancer subtypes, should improve our understanding of their molecular targets. This in turn would facilitate predictions as to their potential usefulness in the clinic.