Flowering time is thought to control fecundity in most natural and agricultural plant species. Selection favoring earlier flowering is commonly observed in the wild; however, it is not clear if plant populations evolve in response to such selection, and why selection for earlier flowering remains detectable. Conversely, why do we not observe selection for later flowering? The authors pose these situations as a paradox, and propose four mechanisms that could result in apparent selection for early flowering, when in fact it is not occurring. The four mechanisms are as follows: 1) fecundity selection is balanced through selection on other fitness components; 2) the fitness function is asymmetrical, making it difficult to measure selection favoring late flowering; 3) condition dependence of plant size, flowering time and fitness; and 4) longer flowering duration, rather than earlier flowering, is favored by selection. (Note: these mechanisms are not mutually exclusive.) However, the authors provide a sturdy foundation for exploring each, and suggest experiments for testing them that should inspire new work in this important area of evolutionary ecology.