Altricial birds often display biased preferences in providing parental care for their dependent offspring, especially during food shortages. During this process, such inflexible rules may result in provisioning errors. To demonstrate how parents optimize their provisioning strategies, we proposed a ‘diagnosis model’ of parental care to posit that parents will undergo a diagnosis procedure to test whether selecting against some particular offspring based on phenotype is an optimal strategy. We tested this model in an asynchronous hatching bird, the Azure-winged Magpie Cyanopica cyanus, based on ten years of data about demography and parental provisioning behaviours. Given their higher daily survival rates, core offspring (those hatched on the first day) merit an investment priority compared with their marginal brood-mates (those hatched on later days). However, a marginal offspring also merited a priority if it displayed greater weight gain than the expected value at the early post-hatching days. Parents could detect such a marginal offspring via a diagnosis strategy, in which they provisioned the brood at the diagnosis stage by delivering food to every nestling that begged, then biased food towards high-value nestlings at the subsequent decision stage by making negative response to the begging of low-value nestlings. In this provisioning strategy, the growth performance of a nestling became a more reliable indicator of its investment value than its hatching order or competitive ability. Our findings provide evidence for this ‘diagnosis model of parental care’ wherein parents use a diagnosis method to optimize their provisioning strategy in brood reduction.