Despite decades of research on management tactics for the navel orangeworm, Amyelois transitella (Walker) (Lepidoptera: Pyralidae), on almonds, we still do not have an established means of using in-season pest-density estimates to predict damage to nuts at harvest. As a result, hull-split pesticide applications, although timed carefully to coincide with navel orangeworm oviposition and with crop vulnerability, are not tied to pest densities—thus falling short of our goals under modern pest management. Here we use an ecoinformatics approach, analyzing a pre-existing data set collected in commercial almond production in California, to ask: 1) are navel orangeworm density estimates obtained using different sampling methods in strong agreement with one another? and 2) can we use either single density estimates or combinations of density estimates to explain variation in nutmeat damage at harvest? We find that correlations between density estimates of navel orangeworm made over a single growing season are often weak, and suggest that density estimates taken closer to the time of harvest (catches of adult females between hull split and harvest; infestation of early-split nuts) may be most useful for predicting damage at harvest. Single-density estimates explained ≤39.1% of variation in harvest damage, whereas a combination of predictors explained 51.5% of the total variance in nutmeat damage at harvest. Our results suggest that density estimates taken just prior to harvest may, with refinement, be usable within a predictive framework to guide late-season control decisions.