We critically examine the procedural deficit hypothesis (PDH) that proposes that a deficit in procedural (as opposed to declarative) learning underlies dyslexia and other developmental disorders. We first note that the existence of dissociated learning disorders (and multiple forms for each disorder) appears incompatible with a general deficit account. Moreover, the PDH formulation appears generally underspecified in terms of predictions to be tested. A particular focus is on the conceptualization of automatization. However, there are alternative views of automaticity, and comparing these different views helps frame the body of findings on the PDH. The insufficient PDH specification led to tasks touching on different skills and selecting target groups based on general diagnostic categories. Accordingly, several recent reviews and meta-analyses reported mixed patterns of findings and reached contradictory conclusions on the PDH. We propose avenues for future research to effectively examine the role of PDH in learning and other developmental disorders.