We briefly review current methods for detecting nonrandom patterns in the temporal overlap of flowering and fruiting curves. We discuss the assumptions behind these methods and propose a new method of analysis using computer simulations to measure n-wise, rather than pairwise, temporal overlap. We quantify the extent to which observed n-wise overlap differs from minimum possible n-wise overlap and apply our method to several data sets to test the hypothesis that interspecific competition for animal visitors has produced flowering curves whose overlap is less than that expected by chance. Results of our analyses provide little support for this hypothesis but suggest alternate criteria by which the species may be selected.