Prenatal exposure of the brain to environmental insult causes different neurological symptoms and behavioral outcomes depending on the time of exposure. To examine the cellular bases for these differences, we exposed Rhesus macaque fetuses to x-rays during early gestation (E30–E42), i.e., before the onset of corticogenesis, or in midgestation (E70–E81) when superficial cortical layers are generated. Animals were delivered at term (~E165), and the size and cellular composition of prefrontal association cortex (area 46) examined in adults, using magnetic resonance imaging (MRI) and stereologic analysis. Both early and midgestational radiation exposure diminished the surface area and volume of area 46. However, early exposure spared cortical thickness and did not alter laminar composition, and due to higher cell density, neuron number was within the normal range. In contrast, exposure to x-rays at midgestation reduced cortical thickness, mainly due to elimination of neurons destined for the superficial layers. A cell-sparse gap, observed within layer III, was not filled by the later generated neurons destined for layer II, indicating that there is no subsequent replacement of the lost neurons. The distinct areal and laminar pathology consequent to temporally segregated irradiation is consistent with basic postulates of the radial unit hypothesis of cortical development. In addition, we show that an environmental disturbance inflicted in early gestation can induce subtle cytoarchitectonic alterations without loss of neurons, such as those observed in schizophrenia, whereas midgestational exposure causes selective elimination of neurons and cortical thinning as observed in some forms of mental retardation and fetal alcohol syndrome.