One-day-old chicks spontaneously peck at a shiny chrome bead. If the bead is coated with methylanthranilate, a bitter tasting substance (M-chicks), they peck once and avoid a similar bead subsequently. Control chicks peck, and continue pecking at a bead dipped in water (W-chicks). Twenty-five hours after this one-trial passive avoidance training the brains were fixed and Golgi-impregnated. A class of large, multipolar, projection neurons from the intermediate medial hyperstriatum ventrale (IMHV) of both hemispheres from M- and W-chicks were examined for changes in their spine density and spine shape. An estimate of the true spine number was obtained using the correction formula of Feldman and Peters. M-chicks showed highly significant increases (P less than 0.0001) in spine density of between 89-113% in the left hemisphere, and 37-69% in the right, compared with W-chicks. There was a significant hemispheric asymmetry in W-chicks: the right hemisphere had approximately 47% more spines per micron than the left, and this difference was abolished after training. Following passive avoidance training, significant increases in spine head diameter (by approximately 9%) and decreases in spine stem length (by approximately 17%), with no significant alterations in overall spine length, were observed in the left hemisphere. The mean dendrite lengths were not significantly changed after training, but an asymmetry of this measure in W-chicks (left greater than right) was present in dendrite orders 2 (P less than 0.01) and 3 (P less than 0.02). These results show that spine densities can increase rapidly (within 25 h) following a one-trial passive avoidance training task and that spine shape changes can be found on the same dendrites which also show changes in spine number. The data support the view that dendritic spines are involved in memory formation processes.